Research Paper: Critically Canadian: Canadian Critical ... · 1990s. Canadian CI sales developed slowly at first but then experienced double-digit growth in the early 2000s. While

Research papers do not necessarily represent the views of the Canadian Institute of Actuaries. Members should be familiar with research papers. Research papers do not

constitute Standards of Practice and therefore are not binding. Research papers may or may not be in compliance with Standards of Practice. Responsibility for the manner of

application of Standards of Practice in specific circumstances remains that of the members in the life insurance practice area.

July 2012

Document 212059

Ce document est disponible en français © 2012 Canadian Institute of Actuaries

Individual Living Benefits Experience Subcommittee

Critically Canadian: Canadian Critical Illness

Standalone Base Incidence Tables

Research Paper

360 Albert Street, Suite 1740, Ottawa ON K1R 7X7 613.236.8196 613.233.4552

[email protected] / [email protected] actuaries.ca / actuaires.ca

Memorandum

To: All Fellows, Affiliates, Associates and Correspondents of the Canadian Institute

of Actuaries

From: Marc-André Melançon, Chair

Member Services Council

Marc-André Belzil, Chair Research Committee

Emile Elefteriadis, Chair Individual Living Benefits Experience Subcommittee

Date: July 30, 2012

Subject: Research Paper: Critically Canadian: Canadian Critical Illness Standalone Base Incidence Tables

This research paper derives incidence rates from general population sources for each of the most common or significant claim triggers (impairments/conditions/surgical procedures) found in the typical Canadian individual standalone critical illness insurance contract.

The derived incidence rates represent population level of expected incidence, adjusted for several factors associated with insurance contracts such as contractual (rather than medical) definition of the condition, first-event diagnosis, survival periods and other factors, trended to 2008. The resulting incidence rates are referred to as the 2008 Canadian CI (CANCI) Tables.

Since the 2008 CANCI tables represent general population level of expected incidence, they are differentiated only by illness, attained age, and gender. Therefore they are not intended to reflect insured lives incidence and do not reflect what would be in the absence of insured experience, hypothetical adjustments for the effects of selection and smoking, and product design. The 2008 CANCI tables will be used as the expected incidence basis for a forthcoming morbidity study on Canadian insured lives experience to be released in late 2012.

A spreadsheet containing the final incidence rates for each claim trigger by attained age and gender is available. The spreadsheet shows the sequential development from the preliminary base incidence rates (interpolated and smoothed) to the final incidence rates.

A research paper of this magnitude required the efforts of many dedicated individuals. We would like to extend our sincere gratitude to these contributors:

Subcommittee Members Graham Dixon Charles Philbrook

Emile Elefteriadis Christopher Piper (2006–2010)

Saul Gercowsky (2006–2008) Anke Roman

Dominic Hains (2007+) Maria Semak (2010+)

Ian Jack Banasha Shah

Frédéric Jacques Debra Shelley

Alethea Lyn Catherine Shum-Adams (2006–2009)

Geoffrey Macdonell

Martin Vézina (2009+)

Research Paper July 2012

4

Table of Contents

EXECUTIVE SUMMARY .......................................................................................................................... 6

1. INTRODUCTION ................................................................................................................................ 6

2. MARKET OVERVIEW ....................................................................................................................... 8

2.1 Types of Coverage Available ........................................................................................................ 8

2.1.1 Individual Market .............................................................................................................. 8

2.1.2 Creditor CI ........................................................................................................................ 9

2.1.3 Group CI ........................................................................................................................... 9

2.2 Current Industry Landscape ........................................................................................................ 10

2.2.1 Market Statistics .............................................................................................................. 10

2.2.2 Sales Experience ............................................................................................................. 10

2.2.3 Underwriting ................................................................................................................... 11

2.2.4 Claims Experience .......................................................................................................... 11

3. METHODOLOGY ............................................................................................................................. 12

3.1 Introduction and Scope ............................................................................................................... 12

3.2 Definitions ................................................................................................................................... 13

3.3 Data Sources ............................................................................................................................... 13

3.4 General Approach ....................................................................................................................... 13

3.5 Exclusions from the Definition ................................................................................................... 14

3.6 Trend ........................................................................................................................................... 14

3.7 First-ever Adjustment ................................................................................................................. 14

3.8 Sudden Death Adjustment and 30-day Survival Adjustment...................................................... 14

3.9 Overlap ........................................................................................................................................ 15

3.10 Prevalence Adjustment ............................................................................................................... 15

4. THE BASE TABLES ......................................................................................................................... 15

4.1 Cancer ......................................................................................................................................... 15

4.1.1 Life-threatening Cancer .................................................................................................. 16

4.1.2 Benign Brain Tumour ..................................................................................................... 25

4.1.3 Early Stage Malignant Melanoma ................................................................................... 31

4.1.4 Early Stage Prostate ........................................................................................................ 37

4.1.5 Ductal Carcinoma in Situ ................................................................................................ 44

4.2 Heart ............................................................................................................................................ 48


5

4.2.1 Heart Attack (Acute Myocardial Infarction) ................................................................... 48

4.2.2 Coronary Artery Bypass Graft (CABG) ......................................................................... 57

4.2.3 Coronary Angioplasty ..................................................................................................... 60

4.2.4 Heart Valve Replacement ............................................................................................... 65

4.2.5 Aortic Surgery ................................................................................................................. 69

4.3 Stroke .......................................................................................................................................... 74

4.4 Kidney Failure ............................................................................................................................ 80

4.5 Major Organ Transplant and Major Organ Failure on Waiting List ........................................... 85

4.6 Multiple Sclerosis ....................................................................................................................... 90

4.7 Alzheimer’s Disease ................................................................................................................... 95

4.8 Parkinson’s Disease .................................................................................................................. 101

4.9 Loss of Independent Existence ................................................................................................. 110

4.10 Minor Conditions ...................................................................................................................... 116

4.10.1 Bacterial Meningitis ...................................................................................................... 117

4.10.2 Blindness ....................................................................................................................... 118

4.10.3 Coma ............................................................................................................................. 118

4.10.4 Deafness ........................................................................................................................ 119

4.10.5 Loss of Limbs ............................................................................................................... 119

4.10.6 Loss of Speech .............................................................................................................. 120

4.10.7 Motor Neuron Disease .................................................................................................. 120

4.10.8 Occupational HIV Infection .......................................................................................... 121

4.10.9 Paralysis ........................................................................................................................ 122

4.10.10 Severe Burns ................................................................................................................. 123

4.10.11 Minor Conditions Summary .......................................................................................... 124

5. ADDITIONAL CONSIDERATIONS .............................................................................................. 125


6

EXECUTIVE SUMMARY This research paper derives incidence rates from general population sources for each of the most common or significant claim triggers (impairments/conditions/procedures) found in the typical Canadian individual standalone critical illness insurance contracts.

The derived incidence rates represent population level of expected incidence, adjusted for several factors associated with insurance contracts such as contractual (rather than medical) definition of the condition, first diagnosis, survival periods and other factors trended to 2008. The resulting incidence rates are referred to as the 2008 Canadian CI (CANCI) Tables.

Since the 2008 CANCI tables represent general population level of expected incidence, they are differentiated only by illness, attained age, and gender. Therefore they are not intended to reflect insured lives incidence and thus do not reflect what would be in the absence of insured experience, hypothetical adjustments for the effects of selection and smoking, and product design. The 2008 CANCI tables will be used as the expected incidence basis for a forthcoming morbidity study on Canadian insured lives experience to be released in late 2012.

The report is divided into five sections.

Section 1 is the introduction to the paper.

Section 2 provides an overview of the Canadian critical illness insurance market and product.

Section 3 provides a description of the general methodology used to develop the incidence rates. This covers the process the subcommittee used for taking the raw base incidence rates from general population and clinical data sources, adjusting the incidence rates to better correspond with the CLHIA bench market definition, including only first-time diagnosis rather than recurrent, and a variety of other adjustments.

Section 4 is the heart of the research paper and provides a detailed derivation of the final incidence rates for each of the most significant claim triggers, along with a description of the data used, and the approximations and judgements applied.

Section 5 covers a variety of topics intended to provide those using the tables with some comfort around the reasonableness of the 2008 CANCI tables, drawing on very high-level Canadian insured lives experience and comparisons between UK and Canadian incidence and mortality rates. In addition, a high-level discussion of the additional adjustments that would be required to estimate incidence rates for insured lives and a review of what these adjustments might look like based on a review of recent insured lives experience from the UK.

A spreadsheet containing the final incidence rates for each claim trigger by attained age and gender is included. The spreadsheet shows the sequential development from the preliminary base incidence rates (interpolated and smoothed) to the final incidence rates.

1. INTRODUCTION This research paper derives incidence rates from general population and clinical data sources for each of the most common claim triggers (conditions/impairments/procedures) found in typical Canadian standalone critical illness insurance contracts.

The derived incidence rates represent population level of expected incidence, adjusted for several factors associated with insurance contracts such as contractual definitions of the claim triggers, first diagnosis, survival periods and other factors trended to 2008. The resulting incidence rates are referred to as the 2008 Canadian CI (CANCI) Tables.


7

Since the 2008 CANCI tables represent general population level of expected incidence, they are differentiated only by claim trigger, attained age, and gender. Therefore they are not intended to reflect insured lives incidence and do not reflect what would be in the absence of insured experience, hypothetical adjustments for the effects of selection and smoking, and product design and other insurance elements. The 2008 CANCI tables will be used as the expected incidence basis for a forthcoming morbidity study on Canadian insured lives experience to be released in late 2012.

The research paper is structured along the lines of similar papers developed in the past for other insurance markets, particularly the UK critical insurance market. The paper begins with an overview of the Canadian CI marketplace then discusses the general methodology used to derive the incidence rates for each of the impairments covered by this paper. The heart of the paper is section 4, where a detailed derivation of incidence rates is developed for each claim trigger.

Section 5 covers a variety of topics intended to provide the reader with some comfort around the reasonableness of the 2008 CANCI tables, drawing on very high-level Canadian insured lives experience and comparisons between UK and Canadian incidence and mortality rates. In addition, a high-level discussion of the additional adjustments that would be required to estimate incidence rates for insured lives as well as a commentary on what those adjustments might look like based on a review of recent insured lives experience from the UK.

A spreadsheet containing the final incidence rates for each claim trigger by attained age and gender is included. The spreadsheet shows the sequential development of the rates from the preliminary base incidence rates (interpolated and smoothed) to the final incidence rates.

It is hoped that this research paper provides a solid foundation for the actuarial community to continue research and increase their understanding of critical illness incidence rates. Some innovative approaches and data sources have been used to derive the incidence rates; for example, estimates of the first-ever and prevalence rates for cancer covered in sections 4.1.1.5 and 4.1.1.8 respectively.

A research paper such as this required the efforts of many dedicated individuals. In fact, the subcommittee's first meeting was October 26, 2006, and it is remarkable that most of the original members have remained members throughout the project. The past and present members of the subcommittee are listed below.

Subcommittee Members Graham Dixon Charles Philbrook

Emile Elefteriadis Christopher Piper (2006–2010)

Saul Gercowsky (2006–2008) Anke Roman

Dominic Hains (2007+) Maria Semak (2010+)

Ian Jack Banasha Shah

Frédéric Jacques Debra Shelley

Alethea Lyn Catherine Shum-Adams (2006–2009)

Geoffrey Macdonell

Martin Vézina(2009+)


8

Since this paper is the result of several different authors’ contribution, the reader will find differences in the writing style and minor inconsistencies in terminology and presentation between the various sections and chapters. Hopefully this does not distract the reader from the underlying content.

Finally, special thanks go to the following:

• Gary Mooney for his independent review and comments on an early draft of this paper;

• Josée Gonthier for the French translation, along with subcommittee members Dominic Hains, Frédéric Jacques, and Martin Vézina, who provided technical review; and

• Martin Labarre for the independent review of the French translation.

2. MARKET OVERVIEW Critical illness insurance (CI) in its current form was first introduced in Canada in the early 1990s. Canadian CI sales developed slowly at first but then experienced double-digit growth in the early 2000s. While individual sales have levelled off between 2004 and 2008, creditor and group CI sales have continued to grow. Public awareness of the product is slowly increasing, but there are challenges to sales growth, such as the price of the coverage, underwriting requirements and advisors’ willingness to learn and offer the product.

Similar to many insurance markets around the world, CI products in Canada initially covered only a few core illnesses. Typically the products covered life-threatening cancers, heart attacks (myocardial infarctions) and strokes (cerebrovascular accidents). However, over time active competition among insurers led to an increase in the number of covered conditions.

CI is sold by both life and health insurers through insurance advisors as well as by banks, the latter offering coverage for outstanding balances on mortgages, loans and credit cards. In addition, the group insurance market, including employer, associations and affinity groups, is viewed as a segment with significant growth potential.

2.1 Types of Coverage Available 2.1.1 Individual Market Unlike markets in other countries, CI was introduced in Canada as a standalone coverage (i.e., a benefit is paid independent of any life insurance coverage). While a few companies do offer an accelerated CI benefit on life insurance policies, there are concerns over the tax status of such policies and so the majority of products currently sold offer CI as a separate benefit. The first products in Canada also included a premium rate guarantee for the entire coverage period, a feature still in place today.

In an effort to compete for sales and make the product more attractive to consumers, new Return of Premium (ROP) features were introduced. These features refund all or a portion of the insurance premiums paid if there is no CI claim, upon specific events such as the policyholder’s death, the surrender of the policy or reaching the end of the coverage period. Although these ROP features are expensive, they have significantly contributed to CI


9

insurance sales growth in recent years. Over half1

Interestingly, the focus of competition among insurers has shifted over the years. It was initially driven by the number of covered illnesses, which has peaked at about 25 conditions, and then it shifted to ROP features. More recently it has been focusing on additional partial benefits which pay out minor benefits (often 10 to 25% of the face amount) on conditions which are seen as serious but not “critical”.

of the new business premium in 2009 was derived from products which also had an ROP option.

In addition, insurers consider CI insurance to be a living benefit and have realized policyholders could benefit from value-added services at the time of a claim. As a result, many insurers offer additional services at the time of a claim, such as:

• Independent physician/specialist evaluation; • Referral to the most appropriate specialist given the claimant’s condition; • Coordination of treatment in the Canadian healthcare system; and • Travel arrangements, accommodation and payment guarantees for claimants choosing

to be treated outside Canada.

Typically, these services are managed and administered by a third party specializing in the medical field. It is still too early to conclude whether CI insurance consumers value these services at the time of a claim. However, initial interest at the point of sale is encouraging.

2.1.2 Creditor CI Banks sell CI coverage combined with life insurance to their mortgage applicants. This sale is done without a licensed insurance advisor. The product tends to be simpler than most, with fewer covered conditions and none of the “return of premium” or other ancillary benefits available on the individual policies. Underwriting tends to be non-medical except perhaps at larger amounts. The CI coverage declines over time as the mortgage is paid down and so the insurer’s risk declines accordingly. Furthermore, the rates are not guaranteed.

The creditor channel has done very well in Canada, selling almost twice as many CI coverages as the individual market. As in the UK, consumers are able to see a direct need for the CI when they purchase their mortgage, as they want to ensure they do not lose their homes if they fall ill. The ease of the sales process is also key to their success.

2.1.3 Group CI Critical illness insurance is available as an employee benefit, either provided as part of a company’s mandatory benefit package or on a voluntary basis. When sold on a voluntary basis, certain minimum participation limits may apply to smaller groups. For larger voluntary CI groups, those who elect the optional coverage may be entitled to a no-evidence limit (NEL) such that no underwriting is required for coverage up to the NEL. Normally a pre-existing condition exclusion is included to help minimize adverse selection and keep incidence rates within the expected range.

Group CI policies tend to have smaller face amounts than individual policies. Premium rates are renewable term and non-guaranteed (i.e., yearly renewable rates). Underwriting is usually not required if the coverage is mandatory, whereas if it is voluntary, medical questionnaires are used for underwriting.

1 Munich Re’s 2010 Individual Insurance Survey.


10

In the last few years, group CI sales have experienced significant growth. More and more providers are entering the group critical illness market, selling products ranging from simple four-illness products to more comprehensive and complex 20-plus condition, multiple-benefit products. The marketplace appears to be quite sensitive to not only price, but product features as well.

2.2 Current Industry Landscape In 2008, there were just over 20 companies offering CI insurance products in Canada. These products continue to be sold through various distribution channels including exclusive insurance agents, independent financial advisors, direct marketing (telemarketing, direct mail, Internet), banks, group insurance representatives, etc. The majority of sales are currently generated in the creditor channels.

2.2.1 Market Statistics The following market statistics are based on an industry-wide survey conducted by Munich Re in Canada (note: this survey currently includes individual insurance sales with guaranteed premium only; the most recent edition includes information on sales up to the end of 2009):

• Policies sold in Canada during 2009 covered, on average 25, critical illnesses (the number of illnesses covered varies from as few as three to as many as 30);

• Issue amounts can be as high as $2,000,000, the average face amount sold being around $84,000 in 2009;

• 67% of 2009 new business premiums were derived from products with ROPs (not including ROP on death); and

• The average issue age in 2009 was 34.5, which is lower than in previous years. This is partly attributable to the emerging juvenile market, which represented 8% of the new business face amount sold in 2009.

Source: Munich Re’s 2010 Individual Insurance Survey.

2.2.2 Sales Experience The table below indicates the growth in new individual insurance sales experienced in Canada over the last few years.


11

Source: Munich Re.

2.2.3 Underwriting The underwriting requirements established to evaluate an insurance applicant vary according to the face amount applied for and the distribution channel. The underwriting practices used for CI products sold through direct marketing, banks and group insurance representatives tend to include only a few questions and/or a pre-existing condition exclusion clause (i.e., the diagnosis of a covered illness found to be related to a medical condition which existed prior to the policy being issued would not be eligible to receive an insurance benefit). On the other hand, policies sold through exclusive and independent insurance advisors require more stringent underwriting requirements, often including blood testing and occasionally a medical examination or possibly review of the applicant’s medical records.

2.2.4 Claims Experience The following graph offers a breakdown of the types of illnesses which have received a benefit under CI individual insurance policies. These statistics include all individual policy claims paid in Canada for those companies participating in the survey until the end of 2009. The results are based on just over 4,700 paid claims and it is not surprising to see life-threatening cancers, heart attacks and strokes being the more common types of illnesses seen so far.

$-

$10 000 000

$20 000 000

$30 000 000

$40 000 000

$50 000 000

$60 000 000

$70 000 000

$80 000 000

$90 000 000

$100 000 000

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

New Business Premiums


12

Source: Munich Re’s 2010 Individual Insurance Survey.

3. METHODOLOGY 3.1 Introduction and Scope Until now there has been no publicly available Canadian critical illness incidence table. Such a table would be useful for benchmarking experience studies and for providing a common benchmark against which any other incidence table can be compared. The purpose of this publication is to create such a table.

The incidence rates derived in this paper are not meant to be a best estimate of insured incidence, nor do they represent pure general Canadian population incidence estimates. The incidence rates account for some of the features of an insured scenario (such as adjustments for the 30-day survival period and overlap between covered conditions) but make no adjustment for others. More specifically there is:

• Only a specific and finite list of covered conditions; • No recognition of smoking status; and • No adjustment for insured selection/underwriting impact (other than prevalence and first-

ever adjustments).

We intend to complete periodic Canadian critical illness experience studies to give clearer indications of how historical insured Canadian incidence rates compare to the rates published here.

We adjusted for overlap of covered conditions. Although this creates a more realistic overall incidence rate, it means that each set of separate covered condition rates have been adjusted under the assumption that some incidences of the illness are included in other covered conditions. Use of the separate covered condition incidence rates therefore requires caution.

All covered conditions have estimates for age last 18 through 80, with some having additional ages included as the data permitted.

68%

13%

5%

3% 4%

7%

Critical Illness Claims Since Inception - % Paid

Cancer

Heart Attack

Stroke

MS

Coronary Bypass

Other Conditions


13

3.2 Definitions Critical illness definitions can vary by insurance carrier and by generation of the product. For the purposes of this paper we selected a single definition for each covered condition and all incidence calculations were based on that definition. The definition is clearly articulated in each covered condition section. To the extent the resulting incidence estimates are used to estimate incidence for critical illness insurance, care must be taken to ensure any differences in definitions are accounted for.

3.3 Data Sources Each covered condition section indicates the data source(s). We used Canadian data wherever possible, although occasionally we reverted to data from other countries.

Four sources of data in particular formed the basis for much of the analysis:

1. Canadian Institute for Health Information (CIHI) is an independent, not-for-profit organization that provides essential data and analysis on Canada’s health system and the health of Canadians. CIHI tracks data in many areas, with information supplied by hospitals, regional health authorities, medical practitioners and governments. Much of CIHI’s data is available freely on its website, but occasionally we made special requests to further refine the data. We used CIHI for some of the key covered conditions, including heart attack and stroke.

2. Statistics Canada is Canada’s central statistical agency, producing statistics on population, resources, economy, society and culture. We used Statistics Canada primarily for the population estimates (i.e., exposures) and also for cancer incidence through its Canadian Cancer Registry.

3. Institute for Clinical and Evaluative Studies (ICES). 4. Canadian Cancer Statistics.

3.4 General Approach The determination of incidence rates for each covered condition followed the same general approach, with any deviations noted in the various covered condition sections:

1. Determination of a definition; 2. Collection of raw data on incidence and exposure; 3. Exclusion of any incidence not aligned with the definition; 4. Trending of the data to 2008; 5. Adjustment to remove multiple incidences of the same covered condition, leaving only

the first-ever incidence for an individual; 6. Removal of individuals who died suddenly from the covered condition—this adjustment

is necessary because these individuals would not have survived long enough to be represented in the 30-day survival adjustment;

7. An adjustment for any overlap with other covered conditions, thus avoiding double-counting of incidence;

8. An exposure adjustment to reflect the prevalence of the illness already present in the exposure population prior to the incidence observations in the data;

9. An adjustment to reflect the 30-day survival period; and


14

10. A summary of all adjustments and the final incidence rates.

In many of the steps in the process we used various smoothing and interpolation techniques. These techniques are noted in the covered condition sections.

3.5 Exclusions from the Definition We tried to acquire raw incidence data that aligned with the definition selected for each covered condition. However, where necessary, we adjusted the raw data to remove incidences that would not have qualified under the definition.

3.6 Trend The incidence rates published represent the incidence rates as at January 1, 2008. They do not represent the incidence rate over an annual period. Therefore each covered condition incidence rate required trending forward to January 1, 2008. Typically we determined the trend using multiple years of the same incidence data. Specific deviations from this approach are noted in the covered condition sections.

3.7 First-ever Adjustment We assumed only the first-ever incidence of a covered condition results in a paid critical illness claim. Many critical illness contracts explicitly state this, and underwriting typically excludes any possibility that subsequent incidences of the same covered condition result in a paid claim.

Many of the data sources for covered conditions include all incidences (i.e., first-ever and recurrent). We adjusted to remove all but the first-ever incidence.

3.8 Sudden Death Adjustment and 30-day Survival Adjustment Individuals diagnosed with a covered condition generally must survive a 30-day waiting period before being eligible for payment of claim. The data available to determine the adjustment for this were typically segmented into those who die suddenly from their illness (e.g., a heart attack victim that dies immediately) and those who survive the initial onset of the illness but die before the 30-day waiting period expires (e.g., a heart attack victim that dies a few days after being admitted to hospital). We considered both adjustments for each covered condition.

Since many conditions do not normally lead to death within the 30-day survival period, we adjusted the incidence rates to reflect the mortality of normal lives.

For this “standard approach” we developed 30-day survival rates based on the results of the 2003–2004 Canadian Standard Ordinary Life Experience Study. We derived age-nearest ultimate mortality rates by applying the overall actual to expected ultimate experience ratio based on amount to the CIA 86–92 aggregate age-nearest mortality rates. We then derived age-last ultimate mortality rates from the age-nearest ultimate mortality rates using the following formula:

AL(x) = [AN(x) + AN(x+1)]/2, where AL(x) = the age-last mortality rate at attained age x; and AN(x) = the age-nearest mortality rate at attained age x.

We determined the 30-day survival rates using the following formula:

30-day survival rate = (1 - AL(x))1/12.


15

3.9 Overlap Many of the covered conditions have material overlap with each other. Examples include:

• An individual can experience any combination of heart attack, coronary artery bypass graft surgery and angioplasty;

• An individual receiving a kidney transplant as a result of kidney failure would satisfy the definition of both kidney failure, major organ transplant, and major organ transplant on waiting list; and

• An individual receiving a bone marrow transplant to treat cancer would satisfy both the cancer and major organ transplant definition.

In each case of overlap, it was necessary to choose one covered condition where the incidence would be included while the other covered conditions would exclude it. Therefore, use caution in using the incidence rates at the covered condition level.

3.10 Prevalence Adjustment To calculate the first incidence of a covered condition, we adjusted the exposures (quite often the Statistics Canada Canadian population estimate) to remove those that already had the covered condition. For example, if the data suggest the incidence is 5% of the Canadian population, it is important to know how many Canadians already had the covered condition. If nobody did, the first-ever incidence rate is 5%. If 25% of the population already had the covered condition, the first-ever incidence is 5% / (1-25%) = 6.7%. In other words, 6.7% of healthy exposures will develop the covered condition.

4. THE BASE TABLES 4.1 Cancer The following five sections cover the cancer-related critical illness covered conditions. Section 4.1.1 addresses life-threatening cancer. By definition, life-threatening cancer excludes benign brain tumour, early stage malignant melanoma, early stage prostate cancer, and ductal carcinoma in situ. These four non-life threatening cancers are the subject of sections 4.1.2, 4.1.3, 4.1.4 and 4.1.5 respectively. Note that the data used to develop the life-threatening cancer incidence rates came from the Canadian Cancer Registry and include all types of invasive cancer sites except non-melanoma skin cancer (basal cell and squamous cell carcinoma of the skin), benign tumors and carcinoma in situ (except for bladder cancer). This means that of the four non-life threatening cancers, early stage malignant melanoma and early stage prostate cancer were included in the data but benign brain tumour and ductal carcinoma in situ were excluded from the data. In developing the life-threatening cancer incidence rates, we decided not to carve out the incidence for early stage malignant melanoma and early stage prostate cancer because, due to data limitations, determinations by stage were fairly subjective and it was conservative to leave these excluded cancers in. In addition, the age-standardized incidence of early stage malignant melanoma and early stage prostate cancer are small relative to all life-threatening cancers—early stage malignant melanoma only accounts for about 1.5% of all life-threatening cancers and early stage prostate cancer accounts for less than 1% of all life-threatening cancers on males. These issues will be discussed in more detail in the sections below.


16

4.1.1 Life-threatening Cancer 4.1.1.1 Definition For the purpose of setting current incidence rates, the Canadian Life and Health Insurance Association (CLHIA) Benchmark Definition captures the relevant features of the Canadian individual critical illness insurance market’s definitions for life-threatening cancer. The CLHIA Benchmark Definition is as follows:

• Cancer (life-threatening) is defined as a definite diagnosis of a tumour characterized by the uncontrolled growth and spread of malignant cells and the invasion of tissue. The diagnosis of Cancer must be made by a Specialist.

• A specialist is a licensed medical practitioner who has been trained in the specific area of medicine relevant to the covered critical illness condition for which benefit is being claimed, and who has been certified by a specialty examining board. In the absence or unavailability of a Specialist, and as approved by the insurer, a condition may be diagnosed by a qualified medical practitioner practising in Canada or the U.S.

Specialist includes, but is not limited to, cardiologist, neurologist, nephrologists, oncologist, ophthalmologist and burn specialist.

Exclusions We have removed the following conditions and dealt with them separately in sections 4.1.2 through 4.1.5:

• Benign brain tumour; • Stage 1A malignant melanoma (melanoma less than or equal to 1.0mm in

thickness, not ulcerated and without Clark level IV or level V invasion); • Any non-melanoma skin cancer that has not metastasized; • Carcinoma in situ; and • Stage A (T1a or T1b) prostate cancer.

Moratorium Period Exclusion No benefit will be payable under this condition if, within the first 90 days following the later of:

• The effective date of the policy, or • The effective date of the last reinstatement of the policy,

the insured person has any of the following:

• Signs or symptoms of cancer, or • Investigations that lead to a diagnosis of cancer (covered or excluded under the

policy), regardless of when the diagnosis is made, or • A diagnosis of cancer (covered or excluded under the policy).

A claimant must report this medical information to the company within six months of the date of diagnosis. If the claimant does not provide this information, the company has the


17

right to deny any claim for cancer or any critical illness caused by any cancer or its treatment.

4.1.1.2 Data Sources 1. Canadian Cancer Statistics, 2007. 2. Custom-made reports prepared by Stats Canada on tables presented in Canadian

Cancer Statistics, 2007. 3. Canadian Standard Ordinary Life Experience—Mortality Study, 2003–2004.

4.1.1.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates The raw observed incidence rates are based on Table 10, Page 50 of the annual report, Canadian Cancer Statistics, 2007, produced by the Canadian Cancer Society, National Cancer Institute of Canada, Statistics Canada, provincial/territorial cancer registries and the Public Health Agency of Canada.

Cancer data collected by the provincial/territorial cancer registries are reported to the patient-oriented Canadian Cancer Registry (CCR) maintained by Statistics Canada. The statistics include all types of invasive cancer sites except non-melanoma skin cancer (basal cell and squamous cell carcinoma of the skin), benign tumors and carcinoma in situ (except for bladder cancer).

Statistics Canada counts only newly diagnosed cases of cancer among people who reside in a given province/territory at the time of diagnosis. The CCR is internally linked to track patients with tumours diagnosed in more than one province/territory in order to reduce duplication. The 2007 new cases are estimates based on 2003 actual data projected forward to 2007, using past numbers of cancers and trends.

The Census and Demographics Branch, Statistics Canada, provided the 2007 population estimates.

Age and Sex Distribution of Cancer

Distribution of All Cancers Combined by Age Group and Gender, Canada, 2007

2007 New Case Estimates 2007 Population Estimates (in 1,000s)

Raw incidence Rates (per 1,000) Age

Group Male Female Total Male Female Total Male Female Total 0-19 700 500 1,200 3,973 3,784 7,757 0.1762 0.1321 0.1547

20-29 850 1,000 1,850 2,292 2,217 4,509 0.3709 0.4511 0.4103 30-39 1,550 2,800 4,350 2,310 2,275 4,585 0.6710 1.2308 0.9487 40-49 4,500 8,300 12,800 2,675 2,660 5,335 1.6822 3.1203 2.3993 50-59 13,300 15,500 28,800 2,244 2,296 4,540 5.9269 6.7509 6.3436 60-69 23,500 17,200 40,700 1,446 1,523 2,969 16.2517 11.2935 13.7083 70-79 24,300 17,100 41,400 886 1,050 1,936 27.4266 16.2857 21.3843 80+ 14,000 14,800 28,800 426 766 1,192 32.8638 19.3211 24.1611

Total 82,700 77,200 159,900 16,252 16,571 32,823 5.0886 4.6587 4.8716 Source: Table 10, Canadian Cancer Statistics, 2007, Page 50


18

The raw incidence rates are defined as the number of new cases of cancer per 100,000 persons in the population. We interpolated/extrapolated these raw incidence rates over the full age range using a three-point Lagrange interpolation formula.

Adjustment for Cancers Excluded from the Definition To qualify for the definition of life-threatening cancer, the tumour must be characterized by the “uncontrolled growth and spread of malignant cells and the invasion of tissue”. As already mentioned, the cancer statistics reported in section 4.1.1.2 include all types of invasive cancer sites except non-melanoma skin cancer (basal cell and squamous cell carcinoma of the skin), benign tumors and carcinoma in situ (except for bladder cancer). Therefore, in order for these cancer statistics to conform to the definition of life-threatening cancer, the following three types of cancers should be removed:

1. In-situ bladder cancer; 2. Stage 1A malignant melanoma (i.e., melanoma less than or equal to 1.0mm in

thickness, not ulcerated and without Clark level IV or level V invasion); and 3. Stage A (T1a or T1b) prostate cancer.

In the end, we made no adjustment to remove the cancers excluded from this definition because:

1. In-situ bladder cancer usually progresses into a more invasive form which is covered by the definition; and

2. Separating out different cancer types by stage is subjective because, until 2008, provinces and territories were not required to report Canadian cancer data by stage and there is not much actual incidence experience by stage to use. Although incidence rates for stage 1A malignant melanoma and stage A (T1a and T1b) prostate cancer have been estimated (see section 4.1.3 and section 4.1.4), for conservatism these incidence rates were not carved out from the incidence rates for life-threatening cancer.

4.1.1.4 Trends To separate the true rate change for cancer incidence from any distortions due to changes in the age distribution of the population or due to population growth, we used age-standardized incidence rates to determine the trend factors. In the 2007 Canadian Cancer Statistics report, the age-standardized incidence rates use the 1991 Canadian population as the standard population.

As mentioned in 4.1.1.3, the 2007 new cases are based on 2003 actual data projected forward to 2007. These projections are calculated by a large panel of statistical experts at Statistics Canada and the Public Health Agency of Canada. We decided that a continuation of the same trend from 2003 to 2007 would be a reasonable estimate of the trend from 2007 to 2008.

The 2007 Canadian Cancer Statistics report that the cancer trend varied by age group and gender, and Statistics Canada provided the actual age-standardized incidence rates by age group and gender for the calendar years 1979 to 2003 and the corresponding projected age-standardized incidence rates for 2004 to 2007. We used these age-standardized incidence rates to develop the annual trend factors from 1980 to 2007. We made the 2008 annual trend factor equal to 2007 estimated trend factor. Since this approach would trend


19

the rates to July 1, 2008, we have backed out six months of trend to take the data back to January 1, 2008, by taking the square root of the 2008 annual trend factor. The following table is an excerpt of the results from this special report:

Development of 2008 Trend Adjustment Factor Trend Factors by Age Group—Males, Canada, 2003

Year Age Group Total 0–19 20–29 30–39 40–49 50–59 60–69 70–79 80+

1994 92.4% 98.9% 105.9% 98.7% 101.3% 99.1% 95.6% 95.8% 97.7% 1995 102.1% 103.3% 97.5% 101.1% 96.0% 93.6% 93.9% 96.0% 95.0% 1996 100.8% 93.5% 100.7% 98.0% 99.8% 98.0% 97.8% 97.8% 98.2% 1997 99.0% 108.2% 90.9% 96.0% 100.8% 101.1% 102.0% 99.9% 100.7% 1998 102.7% 94.8% 99.6% 99.9% 100.0% 100.0% 98.6% 101.9% 99.8% 1999 110.5% 102.5% 98.1% 104.3% 102.8% 104.0% 102.0% 99.6% 102.3% 2000 91.8% 104.0% 97.1% 100.1% 100.6% 100.6% 102.0% 101.7% 101.1% 2001 96.8% 97.0% 100.5% 100.6% 103.5% 102.5% 98.6% 98.0% 100.3% 2002 102.4% 106.3% 101.8% 97.7% 97.4% 95.6% 96.3% 95.7% 96.5% 2003 103.1% 93.5% 97.6% 102.5% 99.0% 98.5% 98.8% 99.1% 98.9% 2004 96.8% 102.0% 106.1% 96.0% 97.5% 98.5% 99.6% 97.7% 98.6% 2005 102.1% 101.2% 98.9% 100.8% 100.8% 101.7% 101.5% 103.5% 101.7% 2006 100.3% 100.5% 99.9% 99.4% 99.3% 99.5% 99.7% 99.8% 99.6% 2007 100.3% 100.5% 99.9% 99.4% 99.3% 99.5% 99.7% 99.8% 99.6% 2008 100.3% 100.5% 99.9% 99.4% 99.3% 99.5% 99.7% 99.8%

1/1/08 100.1% 100.3% 100.0% 99.7% 99.6% 99.7% 99.9% 99.9%

Since the 2008 trend factors were so level by age group, they were set to be flat over each age range.

4.1.1.5 First-ever Adjustment The Statistics Canada cancer data are based on new cases, i.e., new primary cancer sites by individual. Therefore, it is possible for an individual to be counted twice. This can happen in two ways:

1. An individual can be diagnosed with more than one primary cancer site at the same time; or

2. An individual can be diagnosed with a primary cancer site in one year and then a different primary cancer site in a later year.

To remove the duplication in the cancer data, Statistics Canada redid its 2003 new cases data on a first-ever incidence basis. For this report, any individuals with prior cancers back to 1969 were excluded. The results of this special report are tabulated below:


20

Development of First-ever Incidence Adjustment Factor Cancer Incidence by Age Group and Gender, Canada, 2003

2003 New Cases(1)

2003 Persons with more than

one Primary Cancer Site(2)

2003 Persons with Prior

Diagnosis in 1992–2002(3)

2003 Persons with Prior

Diagnosis in 1969–1991(3)

2003 First-ever Incidence Cases

First-ever Incidence

Adjustment Factor

Male Female Male Female Male Female Male Female Male Female Male Female

0–4 213 150 4 2 2 0 0 0 207 148 97.2% 98.7% 5–9 124 107 0 0 0 0 0 0 121 105 97.6% 98.1%

10–14 127 114 0 0 3 3 0 0 124 111 97.6% 97.4% 15–19 246 207 1 1 3 2 5 1 237 203 96.3% 98.1% 20–24 357 348 4 7 0 2 3 3 350 336 98.0% 96.6% 25–29 428 565 2 0 1 4 1 3 424 558 99.1% 98.8% 30–34 573 1,021 3 5 5 5 5 9 560 1,002 97.7% 98.1% 35–39 933 1,821 8 7 8 17 5 14 912 1,783 97.7% 97.9% 40–44 1,667 3,186 17 13 32 51 17 29 1,601 3,093 96.0% 97.1% 45–49 2,849 4,776 32 45 58 121 24 46 2,735 4,564 96.0% 95.6% 50–54 4,777 5,851 52 57 136 197 41 95 4,548 5,502 95.2% 94.0% 55–59 7,352 6,969 107 83 259 320 87 132 6,899 6,434 93.8% 92.3% 60–64 9,379 7,075 145 100 428 419 131 170 8,675 6,386 92.5% 90.3% 65–69 11,037 7,179 217 98 694 429 188 248 9,938 6,404 90.0% 89.2% 70–74 12,301 8,139 309 128 1,124 539 276 343 10,592 7,129 86.1% 87.6% 75–79 10,498 8,131 258 140 1,127 581 341 385 8,772 7,025 83.6% 86.4% 80–84 6,880 6,550 172 99 813 443 285 365 5,610 5,643 81.5% 86.2% 85+ 4,561 5,708 105 69 482 350 244 375 3,730 4,914 81.8% 86.1%

Total 74,302 67,897 1,436 854 5,175 3,483 1,653 2,218 66,035 61,340 88.9% 90.3%

Sources: 1. Statistics Canada. Cancer Incidence in Canada, 2003 and 2004, Second Edition,

Table 1-1. 2. Statistics Canada. Cancer Incidence in Canada, 2003 and 2004, Second Edition,

Table 2-1. 3. Statistics Canada.

The 2003 new cases are the cancer data as reported and used by Statistics Canada. These data are adjusted for persons who were diagnosed with more than one primary cancer site in 2003 and persons diagnosed with a different primary cancer site prior to 2003. We determined the 2003 first-ever incidence cases using the following formula:

2003 First-ever Incidence Cases = 2003 New Cases Less: 2003 Persons with more than one Primary Cancer Site Less: 2003 Persons with Prior Diagnosis in 1992–2002 Less: 2003 Persons with Prior Diagnosis in 1969–1991.


21

We determined the First-ever Incidence Adjustment Factor using the following formula:

First-ever Incidence Adjustment Factor = 2003 First-ever Incidence Cases / 2003 New Cases.

We interpolated/extrapolated these first-ever incidence adjustment factors over the full age range using a three-point Lagrange interpolation formula.

4.1.1.6 Sudden Death Adjustment Provincial and territorial registries maintain death records for people residing in that province or territory at the time of death. These records are used to identify cancer deaths, i.e., deaths attributed to some form of cancer as the underlying cause of death by the certifying physician. In general, cancer deaths occurring in a given year will usually be the result of cancers diagnosed in previous years. Therefore, we made no adjustment.

4.1.1.7 Overlap Adjustment There is no firm evidence that the existence of the other studied critical illness covered conditions affects the incidence of life-threatening cancer. Therefore, we made no adjustment for overlap with other critical illnesses.

4.1.1.8 Prevalence Adjustment The 2007 Statistics Canada Population Estimate used in section 4.1.1.3 must be adjusted to remove all individuals who have previously been diagnosed with life-threatening cancer.

The 2007 Canadian Cancer Statistics report provided a prevalence count for the number of Canadians who were alive in 2003 within 15 years of their cancer diagnosis. These prevalence counts were calculated by applying survival rates from a Saskatchewan study to the Canadian annual incidence numbers.

However, it is possible for individuals who have had cancer to survive much longer than 15 years. So we used Statistics Canada’s background data on its Saskatchewan study to calculate a complete prevalence estimate.


22

Development of Prevalence Adjustment Factor Prevalence Counts by Age Group—Males, Canada, 2003

Age Group

Number of Years of Survival 2003 Population

Prevalence Adjustment

Factor 5 10 11 12 13 14 15 Complete

00–04 843 1,010 1,010 1,010 1,010 1,010 1,010 1,010 874,449 0.1%

05–09 534 1,155 1,201 1,234 1,261 1,294 1,322 1,322 1,000,126 0.1%

10–14 538 991 1,109 1,218 1,342 1,463 1,572 1,572 1,086,111 0.1%

15–19 926 1,381 1,453 1,532 1,606 1,683 1,762 2,114 1,090,073 0.2%

20–24 1,318 2,021 2,108 2,189 2,268 2,338 2,411 2,893 1,120,041 0.3%

25–29 1,842 2,884 3,036 3,179 3,317 3,437 3,547 4,256 1,075,917 0.4%

30–34 2,459 4,055 4,254 4,456 4,643 4,840 5,046 6,055 1,127,209 0.5%

35–39 3,026 5,681 6,014 6,354 6,701 7,030 7,347 8,816 1,250,331 0.7%

40–44 5,022 7,325 7,845 8,346 8,767 9,210 9,599 11,519 1,366,129 0.8%

45–49 8,426 11,565 11,923 12,265 12,517 12,765 12,980 15,576 1,252,859 1.2%

50–54 14,344 19,327 19,827 20,271 20,655 21,028 21,337 25,604 1,079,639 2.4%

55–59 21,310 29,087 29,846 30,512 31,084 31,601 32,056 38,467 913,862 4.2%

60–64 28,036 39,389 40,514 41,508 42,320 43,070 43,782 52,538 684,714 7.7%

65–69 33,175 50,314 52,093 53,656 55,103 56,401 57,593 69,112 553,269 12.5%

70–74 35,749 51,017 54,025 56,571 58,811 60,838 62,665 75,198 485,226 15.5%

75–79 30,352 46,120 47,938 49,270 50,374 51,267 51,934 62,321 359,682 17.3%

80–84 18,691 30,965 32,791 34,158 35,237 36,078 36,720 44,064 222,542 19.8% 85+ 12,540 25,369 28,251 30,509 32,378 33,880 35,009 42,011 136,822 30.7%

All 219,172 329,697 345,279 358,279 369,433 379,275 387,734 464,450 15,679,001


23

Development of Prevalence Adjustment Factor Prevalence Counts by Age Group—Females, Canada, 2003

Age Group

Number of Years of Survival 2003 Population

Prevalence Adjustment

Factor 5 10 11 12 13 14 15 Complete

00–04 675 835 835 835 835 835 835 835 834,274 0.1%

05–09 423 997 1,025 1,060 1,095 1,123 1,149 1,149 952,520 0.1%

10–14 498 878 989 1,103 1,199 1,290 1,380 1,380 1,034,099 0.1%

15–19 837 1,253 1,331 1,408 1,472 1,542 1,586 1,903 1,033,182 0.2%

20–24 1,353 1,990 2,071 2,154 2,221 2,276 2,345 2,814 1,070,431 0.3%

25–29 2,391 3,493 3,620 3,734 3,847 3,981 4,083 4,900 1,045,993 0.5%

30–34 4,327 6,370 6,566 6,783 6,979 7,169 7,370 8,844 1,105,438 0.8%

35–39 7,241 11,226 11,643 12,013 12,417 12,834 13,231 15,877 1,235,422 1.3%

40–44 1,197 17,503 18,272 19,028 19,701 20,376 21,012 25,214 1,356,581 1.9%

45–49 1,710 25,297 26,215 27,074 27,838 28,523 29,173 35,008 1,265,207 2.8%

50–54 22,085 33,892 35,224 36,514 37,686 38,766 39,712 47,654 1,098,736 4.3%

55–59 24,308 37,935 39,791 41,443 42,812 44,113 45,219 54,263 928,818 5.8%

60–64 25,101 40,113 42,132 43,921 45,547 47,055 48,400 58,080 712,128 8.2%

65–69 22,728 40,186 42,636 44,805 46,887 48,758 50,543 60,652 596,195 10.2%

70–74 25,097 37,719 40,753 43,599 46,336 48,851 51,135 61,362 555,706 11.0%

75–79 25,074 38,971 40,705 42,142 43,339 44,388 45,302 54,362 481,491 11.3%

80–84 18,924 31,205 32,989 34,450 35,603 36,566 37,396 44,875 361,699 12.4% 85+ 16,390 32,859 36,459 39,270 41,666 43,715 45,487 54,584 313,545 17.4%

All 226,556 362,751 383,284 401,364 417,508 432,188 445,385 533,757 15,981,465

Over a 15-year survival period, the prevalence counts for ages 0–14 should be complete. However, according to the paper “Estimating the Completeness of Prevalence Based on


24

Cancer Registry Data”2

We set the prevalence adjustment factor using the following formula:

, the calculated prevalence of all cancers combined would drop by 16% if based on data for only 17 years. To determine the completeness index for a 15-year survival period, the annual rate of increase in prevalence counts from 12 years to 15 years was studied for ages 15+. The average rate of increase was 3.7% from a 12-year to a 13-year survival period, 3.3% from a 13-year to 14-year survival period and 2.9% from a 14-year to a 15-year survival period. The annual average rate of increase decreased by -0.4% per year. Therefore, the completeness index for a 15-year survival period was estimated to be 16% + (2.9% - 0.4%) + (2.9% - 0.8%) or approximately 20%. To complete the prevalence count, the prevalence count for a 15-year survival period was grossed up by 20% for ages 15+.

Prevalence Adjustment Factor = Complete Prevalence Count / 2003 Population.

We interpolated/extrapolated these prevalence adjustment factors over the full age range using a three-point Lagrange interpolation formula.

4.1.1.9 Thirty-day Survival Adjustment As mentioned in 4.1.1.6, cancer deaths usually result from cancers diagnosed in previous years. Since life-threatening cancer does not normally lead to death within the 30-day survival period we adjusted the incidence rates to reflect the mortality of normal lives. Please refer to section 3.8 for the details on the development of standard mortality.

4.1.1.10 Summary of Rates and Adjustments

Summary of Calculation of Incidence Rates per 1,000: Cancer Males Females Central Age 35 55 75 35 55 75 Base Rate 0.6771 6.4230 27.8348 1.2839 6.9673 16.4840

Adjustments: Cancers Excl. from Definition 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

2008 Trend 0.0% -0.4% -0.1% 0.3% 0.3% 0.2% First-ever Incidence -2.1% -5.6% -15.5% -1.9% -7.0% -13.2%

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.6% 3.3% 16.5% 1.1% 5.1% 11.1% 30-Day Survival 0.0% 0.0% -0.3% 0.0% 0.0% -0.2% Derived Incidence Rate 0.6672 6.2420 28.0591 1.2763 6.8514 16.0788

2 Capocaccia, R., and R. De Angelis, Statistics in Medicine 16 (1997): 425–440.


25

Note: The Derived Incidence Rate for life-threatening cancer = (Base Rate) x (1 + Cancers Excl. from Definition Adjustment) x (1 + 2008 Trend Adjustment) x (1 + First-ever Incidence Adjustment) / (1 - Sudden Death Adjustment) x (1 + Overlap Adjustment) / (1 - Prevalence Adjustment) x (1 + 30-Day Survival Adjustment)

A comparison of the raw and derived incidence rates per 1,000 for ages 15–85 is graphed below.

4.1.2 Benign Brain Tumour

4.1.2.1 Definition For the purpose of determining current incidence rates, we used the following definition to capture the relevant features of the Canadian individual critical illness insurance market’s definitions for benign brain tumour:

Benign brain tumour is defined as a non-malignant tumour located in the cranial vault and limited to the brain, meninges, cranial nerves or pituitary gland. The tumour must require surgery or radiation treatment or cause irreversible objective

Life-Threatening Cancer Incidence Rates Initial vs Final Basis

0.0000

5.0000

10.0000

15.0000

20.0000

25.0000

30.0000

35.0000

40.0000

15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84

Age Last

Rat

e pe

r 1,0

00

Initial Male Initial Female Final Male Final Female


26

neurological deficits. No benefit will be payable under this critical illness insured condition for pituitary adenomas less than 10mm.






A claimant must report this medical information to the company within six months of the date of diagnosis. If the claimant does not provide this information, the company has the right to deny any claim for cancer or any critical illness caused by any cancer or its treatment.

4.1.2.2 Data Sources 1. Cancer Surveillance Online, Public Health Agency of Canada. Cancer of the

Brain, 2002, 2003, and 2004, Canada (extracted June 17, 2010). 2. Kaltas et al. “Diagnosis and Management of Pituitary Carcinomas”. The Journal

of Clinical Endocrinology & Metabolism (March 1, 2005). Website: http://jcem.endojournals.org/cgi/content/full/90/5/3089.

3. Central Brain Tumor Registry of the United States (CBTRUS). CBTRUS Statistical Report: Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2004–2006. Hinsdale, IL, 2010. Website: www.cbtrus.org.

4.1.2.3 Calculation of base incidence rates The Canadian Cancer Registry confirmed that it does not track benign brain tumours in its database. It only tracks malignant brain tumours. However, the U.S. has a Central Brain Tumor Registry which tracks both benign and malignant brain tumours. Based on the U.S. data, we can develop benign to malignant brain tumour ratios and apply these ratios to the Canadian malignant brain tumour incidence rates to estimate benign brain tumour incidence rates in Canada.

This approach assumes that the proportion of benign to malignant brain tumour incidence in Canada is the same as in the U.S. There are no studies to corroborate this assumption. However, as with the development of incidence rates for ductal carcinoma in situ in section 4.1.5, given the relative immateriality of benign brain tumour to total critical illness incidence, this assumption is not unreasonable.

Unadjusted Raw Observed Incidence Rates The actual incidence rates for 2004 were too erratic, so we used the calendar years 2002, 2003 and 2004 incidence data to develop the distribution. The averages of the 2002–2004

http://jcem.endojournals.org/cgi/content/full/90/5/3089�

http://www.cbtrus.org/�


27

actual incidence rates for malignant cancer of the brain produced by Cancer Surveillance Online are tabulated below.

The source of the incidence data is the Canadian Council of Cancer Registries. Data quality information was provided by the provincial and territorial cancer registries and the Health Statistics Division of Statistics Canada. The new cases of cancer of the brain were classified according to the ICD-O-3 Site/Type codes (International Classification of Diseases for Oncology – Third Edition) C710:C719 excluding M-905, 9140, 953, and M-9590:9989.

Population estimates were provided by Statistics Canada.

The raw incidence rates are defined as the number of new cases of cancer per 1,000 persons in the population.

Adjustment from Malignant to Benign Brain Tumour Incidence Rates In 2002, the Benign Brain Tumor Cancer Registries Amendment Act (Public Law 107-206) was passed. This law requires that all state cancer registries must include data on primary non-malignant brain and central nervous system (CNS) tumours, beginning with the 2004 diagnosis year. Prior to this law, benign brain tumours were not consistently reported. For this reason, CBTRUS reported data was limited to the 2004 diagnosis year and later.

The CBTRUS statistical report Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2004–2006 provided the 2004–2006 actual incidence rates per 100,000 for benign and malignant brain tumours, but not in the level of detail

Male Female15-19 1.85 1.5520-24 1.80 1.8825-29 3.07 2.0930-34 3.84 2.7135-39 4.37 3.1140-44 5.14 3.1545-49 8.00 4.4650-54 9.22 6.2055-59 12.77 8.6560-64 18.73 12.1465-69 23.08 13.4770-74 25.21 17.2975-79 29.28 20.1580-84 28.58 20.7485+ 25.42 15.44

All Ages 7.47 5.03

Distribution of Cancer of the Brain Combined by Age Group and Gender, Canada, 2002 - 2004

Age Group

2002 - 2004 Raw Incidence Rates

(per 100,000)


28

that we required for our analysis. We requested a customized query from CBTRUS, breaking down the incidence rates by behaviour, gender and age. CBTRUS cautioned that it could not reproduce the study results because it is not allowed to do further analyses on the entire dataset, but it could run our query on the data from 17 registries (54,497 cases). We compared the query results against the report results and the deviation was less than 4.0%. Based on the query, the actual incidence rates per 100,000 for adults diagnosed with brain tumours in the U.S. in 2004–2006, by behaviour and gender, and the corresponding ratios of benign to malignant incidence rates by gender, are as follows:

It should be noted that the Canadian cancer of the brain incidence rates cover ICD-O-3 codes C710–C719, which cover brain sites only. The U.S. CBTRUS brain and central nervous system tumour incidence rates cover ICD-O-3 codes for C71.0–C71.9 (brain), C70.0–C70.9 (meninges), C72.0–C72.5, C72.8–C72.9 (central nervous system), C75.1–C75.3 (pituitary and pineal glands), and C30.0: 9522–9523 (olfactory tumours). Although the U.S. malignant incidence rate includes more sites than the Canadian malignant incidence rate, brain tumours account for most of the U.S. malignant incidence rate and therefore, the distortion due to the additional sites included is not material.

Adjusted Raw Observed Incidence Rates The U.S. benign to malignant brain tumour incidence ratios, derived in the previous subsection, were applied to the Canadian incidence rates for malignant cancer of the brain to estimate the incidence rates for benign brain tumours in Canada.

We interpolated/extrapolated these adjusted raw observed incidence rates over the full age range using a three-point Lagrange interpolation formula.

Adjustment for Cancers Excluded from the Definition The definition states that “No benefit will be payable under this critical illness insured condition for pituitary adenomas less than 10mm.” Since we do not have any information on the breakdown of benign brain tumour incidence by site and size in Canada, we turned to the U.S. for our analysis. According to the paper “Clinical Review: Diagnosis and Management of Pituitary Carcinomas” (Kaltas et al), most pituitary tumours are “clinically insignificant and less than 5mm in diameter”. According to the CBTRUS statistical report Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2004–2006, pituitary tumours accounted for 2.39 per 100,000 (or 21%) of the non-malignant brain and central nervous system incidence rate of 11.52 per 100,000. We decided to remove 50% of the pituitary tumour incidence rate to account for

Male Female

Malignant Brain Tumour Incidence Rates 10.66 7.41

Benign Brain Tumour Incidence Rates 12.27 19.38

Benign/Malignant Ratio 1.151 2.615

U.S. Brain Tumour Incidence Rates per 100,000 Diagnosed in 2004 - 2006, For ages 20+

(CBTRUS)


29

the exclusion of pituitary tumours less than 10mm in diameter in the definition. We felt that the 50% reduction was conservative given that most pituitary tumours are less than 5mm in diameter. Since pituitary tumours accounted for 21% of the total non-malignant brain tumour incidence rate, we reduced the benign brain tumour incidence rates by 50% x 21% = 10%.

4.1.2.4 Trends According to the CBTRUS statistical report Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2004–2006, the actual incidence rates did not differ by behaviour from 2004 to 2006. The report observes that there are no statistically significant trends in benign or malignant brain tumour incidence rates. Therefore, no adjustment for trend was used.

4.1.2.5 First-ever Adjustment The benign brain tumour incidence rates were derived from the same data source as the life-threatening cancer incidence rates. Therefore, the adjustment for life-threatening cancer was used. See section 4.1.1.5.

4.1.2.6 Sudden Death Adjustment As with life-threatening cancer, no adjustment was made for sudden death. See Section 4.1.1.6.

4.1.2.7 Overlap Adjustment As with life-threatening cancer, no adjustment was made for overlap with other critical illness conditions. See section 4.1.1.7.

4.1.2.8 Prevalence Adjustment The benign brain tumour incidence rates were derived from the same data source as the life-threatening cancer incidence rates. Therefore, the adjustment for life-threatening cancer was used. See section 4.1.1.8.

4.1.2.9 Thirty-day Survival Adjustment As with life-threatening cancer, the mortality for normal lives was used. See section 3.8.


30


Central Age 35 55 75 35 55 75 Base Rate 0.0047 0.0127 0.0325 0.0078 0.0197 0.0504

Adjustments: Definition Adjustment -10.0% -10.0% -10.0% -10.0% -10.0% -10.0% 2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First Ever Incidence -2.1% -5.6% -15.5% -1.9% -7.0% -13.2% Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.6% 3.3% 16.5% 1.1% 5.1% 11.1% 30-Day Survival 0.0% 0.0% -0.3% 0.0% 0.0% -0.2%

Derived Incidence Rate 0.0042 0.0112 0.0295 0.0070 0.0174 0.0442

Note: The Derived Incidence Rate = (Raw Observed Rate) x (1 + Definition Adjustment) x (1 + 2008 Trend Adjustment) x (1 + First Ever Incidence Adjustment) / (1 - Sudden Death Adjustment) x (1 + Overlap Adjustment) / (1 - Prevalence Adjustment) x (1 + 30-Day Survival Adjustment)

Summary of Calculation of Incidence Rates per 1,000: Benign Brain Tumour Males Females


31


4.1.3 Early Stage Malignant Melanoma

4.1.3.1 Definition For the purpose of determining current incidence rates, we used the following definition to capture the relevant features of the Canadian individual critical illness insurance market’s definitions for early stage malignant melanoma:

Early stage malignant melanoma is defined as stage 1A malignant melanoma (melanoma less than or equal to 1.0mm in thickness, not ulcerated and without Clark level IV or level V invasion), as confirmed by biopsy.





policy), regardless of when the diagnosis is made, or

Benign Brain Tumour Incidence Rates Initial vs Final Basis

0.00

0.01

0.02

0.03

0.04

0.05

0.06

15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

Age Last

Rate

per

1,0

00



32

• A diagnosis of cancer (covered or excluded under the policy). A claimant must report this medical information to the company within six months of the date of diagnosis. If the claimant does not provide this information, the company has the right to deny any claim for cancer or any critical illness caused by any cancer or its treatment.

4.1.3.2 Data Sources 1. Cancer Surveillance Online, Public Health Agency of Canada. Malignant

Melanoma of the Skin, 2003, Canada (extracted April 1, 2008). 2. Surveillance, Epidemiology, and End Results (SEER) Program. SEER*Stat

Database: Incidence—SEER 17 Regs Limited-Use + Hurricane Katrina Impacted Louisiana Cases, Nov 2007 Sub (1973–2005 varying)—Linked To County Attributes—Total U.S., 1969–2005 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch released April 2008, based on the November 2007 submission. Website: www.seer.cancer.gov.

4.1.3.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates The raw observed incidence rates are based on the 2003 actual incidence rates produced by Cancer Surveillance Online.

http://www.seer.cancer.gov/�


33

The source of the incidence data is the Canadian Council of Cancer Registries. Data quality information was provided by the provincial and territorial cancer registries and the Health Statistics Division of Statistics Canada. The new cases of malignant melanoma of the skin were classified according to the ICD-O-3 Site/Type codes (International Classification of Diseases for Oncology – Third Edition) C440:C449 (types 8720:8790). This includes all types of malignant melanoma skin cancers and excludes non-melanoma skin cancer (basal cell and squamous cell carcinoma of the skin).

Population estimates were provided by Statistics Canada.

The raw incidence rates are defined as the number of new cases of cancer per 1,000 persons in the population.

We interpolated/extrapolated these raw incidence rates over the full age range using a three-point Lagrange interpolation formula.

Adjustment for Cancers Excluded from the Definition To qualify for the definition of early stage malignant melanoma, the malignant melanoma must be stage 1A. Stage 1A is defined as lesions less than or equal to 1.0mm in thickness, with no evidence of ulceration, and Clark Level II–III (T1aN0M0). Note that the

Distribution of Malignant Melanoma of the Skin Combined by Age Group and Gender,

Canada 2003 2003 Actual New

Cases 2003 Raw

Incidence Rates (per 1,000) Age

Group Male Female Male Female 0-4 0 0 0.0000 0.0000 5-9 0 0 0.0000 0.0000

10-14 0 0 0.0000 0.0000 15-19 8 13 0.0073 0.0126 20-24 30 43 0.0268 0.0402 25-29 43 77 0.0399 0.0736 30-34 62 105 0.0549 0.0949 35-39 86 112 0.0687 0.0906 40-44 138 170 0.1009 0.1253 45-49 188 182 0.1500 0.1439 50-54 224 182 0.2075 0.1657 55-59 237 169 0.2594 0.1820 60-64 233 134 0.3404 0.1883 65-69 238 140 0.4299 0.2347 70-74 239 147 0.4923 0.2643 75-79 196 135 0.5443 0.2803 80-84 126 116 0.5670 0.3208 85+ 92 96 0.6723 0.3066

Total 2,140 1,821 0.1274 0.0997


34

definition states that the melanoma must be without Clark Level IV or Level V invasion. The Clark level describes the melanoma according to the layers of skin involved. Clark Levels II and III indicate that the melanoma has spread to the upper dermis. Clark Level IV indicates that the melanoma has spread to the lower dermis and Clark Level V indicates that the melanoma has spread to the subcutis. As already mentioned in the above section on raw observed incidence rates, the cancer statistics reported include all malignant melanomas. Therefore, in order for these cancer statistics to conform to the definition of early stage malignant melanoma, the stage 1A melanomas need to be separated out.

The Canadian cancer registries are just in their infancy in collecting data by stage and we could not access any credible Canadian experience to determine the proportion of malignant melanomas attributable to stage 1A. We turned to the U.S. SEER database which has the appropriate staging breakdown for calendar years of diagnosis 2004 and 2005 (see table below). Based on this U.S. experience, we determined that the proportion of U.S. malignant melanomas attributable to stage 1A was 54%. We applied this ratio to the Canadian malignant melanoma experience to estimate the proportion of Canadian stage 1A cases.

4.1.3.4 Trends To separate the true rate change for cancer incidence from any distortions due to changes in the age distribution of the population or due to population growth, we used age-standardized incidence rates to determine the trend factors. In the Cancer Surveillance Online reporting, the age-standardized incidence rates use the 1991 Canadian population as the standard population.

The age-standardized incidence rates per 100,000 for actual Canadian malignant melanoma experience from 1992 to 2004 are tabulated below. The average annual trend

Distribution of Malignant Melanoma of the Skin Combined by AJCC Stage,

U.S. SEER 17 Database, 2004 - 2005 Unadjusted Stage

Distribution Adjusted Stage

Distribution (removed Unknown) AJCC

Stages Count Percentage Count Percentage IA 9,724 43.2% 9,724 54.0% IB 3,962 17.6% 3,962 22.0% IIA 1,129 5.0% 1,129 6.3% IIB 748 3.3% 748 4.2% IIC 320 1.4% 320 1.8%

IIINOS 138 0.6% 138 0.8% IIIA 421 1.9% 421 2.3% IIIB 502 2.2% 502 2.8% IIIC 285 1.3% 285 1.6% IV 783 3.5% 783 4.3%

Unknown 4,512 20.0% Total 22,524 100.0% 18,012 100.0%


35

factor to apply to the 2003 raw observed incidence rates to project them forward to 2008 was determined by averaging the actual annual trend from 1992 to 2004. Since this approach would trend the rates to July 1, 2008, six months of trend were backed out to take the data back to January 1, 2008, by taking the square root of the 2008 annual trend factor. This resulted in a 2003 to 2008 trend factor of 98.2% x (101.6%)3.5 = 103.8% for males and 102.7% x (101.4%)3.5 = 107.8% for females.

4.1.3.5 First-ever Adjustment The early stage malignant melanoma incidence rates were derived from the same data source as the life-threatening cancer incidence rates. In addition, since malignant melanoma is more common at older ages, it is possible for an individual to contract another cancer, recover and then get malignant melanoma. In fact, the chemotherapy and radiation treatment for some cancers (e.g., Hodgkin’s lymphoma) may increase the likelihood of the treated individual getting malignant melanoma. Due to data limitations the adjustment for life-threatening cancer was used. See section 4.1.1.5.

4.1.3.6 Sudden Death Adjustment As with life-threatening cancer, no adjustment was made for sudden death. See section 4.1.1.6.


Malignant Melanoma of the Skin Age-Standardized Incidence Rates per 100,000,

Canada (1991)

Age-standardized Incidence Rates

Annual Trend

Male Female Male Female 1992 10.33 8.67 1993 10.30 8.94 99.7% 103.1% 1994 10.68 9.07 103.7% 101.5% 1995 11.08 9.28 103.7% 102.3% 1996 10.97 9.52 99.0% 102.6% 1997 11.29 9.50 102.9% 99.8% 1998 10.95 9.53 97.0% 100.3% 1999 12.67 9.99 115.7% 104.8% 2000 12.45 10.28 98.3% 102.9% 2001 12.63 10.17 101.4% 98.9% 2002 12.04 9.85 95.3% 96.9% 2003 12.74 9.97 105.8% 101.2% 2004 12.51 10.24 98.2% 102.7% 2005 12.71 10.38 101.6% 101.4% 2006 12.92 10.53 101.6% 101.4% 2007 13.12 10.68 101.6% 101.4%

1/1/2008 13.33 10.82 100.8% 100.7%


36

4.1.3.8 Prevalence Adjustment The early stage malignant melanoma incidence rates were derived from the same data source as the life-threatening cancer incidence rates. Therefore, the adjustment for life-threatening cancer was used. See section 4.1.1.8.

4.1.3.9 Thirty-Day Survival Adjustment As with life-threatening cancer, the mortality for normal lives was used. See section 3.8.



Central Age 35 55 75 35 55 75 Base Rate 0.0610 0.2351 0.5270 0.1113 0.1767 0.2710

Adjustments: Definition Adjustment -46.0% -46.0% -46.0% -46.0% -46.0% -46.0% 2008 Trend 3.8% 3.8% 3.8% 7.8% 7.8% 7.8% First Ever Incidence -2.1% -5.6% -15.5% -1.9% -7.0% -13.2% Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.6% 3.3% 16.5% 1.1% 5.1% 11.1% 30-Day Survival 0.0% 0.0% -0.3% 0.0% 0.0% -0.2%


Note: The Derived Incidence Rate = (Raw Observed Rate) x (1 + Definition Adjustment) x (1 + 2008 Trend Adjustment) x (1 + First Ever Incidence Adjustment) / (1 - Sudden Death Adjustment) x (1 + Overlap Adjustment) / (1 - Prevalence Adjustment) x (1 + 30-Day Survival Adjustment)

Summary of Calculation of Incidence Rates per 1,000: Early Stage Malignant Melanoma Males Females


37

4.1.4 Early Stage Prostate

4.1.4.1 Definition For the purpose of setting current incidence rates, the CLHIA Benchmark Definition captures the relevant features of the Canadian individual critical illness insurance market’s definitions for Early Stage Prostate Cancer. The CLHIA Benchmark Definition is as follows:

Early stage prostate is defined as stage A (T1a or T1b) prostate cancer.

Stage A (T1a/T1b) prostate cancer is classified according to the following ICD-O-3 Site/Type codes (International Classification of Diseases for Oncology – Third Edition): (ICDO-3 C619 excluding M-905, 9140, 9590:9989).




• Signs or symptoms of cancer, or • Investigations, that lead to a diagnosis of cancer (covered or excluded under

the policy), regardless of when the diagnosis is made, or • A diagnosis of cancer (covered or excluded under the policy).

Early Stage Malignant Melanoma Incidence Rates Initial vs Final Basis

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

Age Last

Rate

per

1,0

00



38

* Stage A (T1a/T1b) Prostate Cancer is classified according to the following ICD-O-3 Site/Type codes (International Classification of Diseases for Oncology - 3rd Edition): (ICDO-3 C619 excluding M-905, 9140, 9590:9989).

4.1.4.2 Data Sources 1. Canadian Cancer Statistics, 2007. 2. CANSIM table 103-0550, Canadian Cancer Registry (CCR) Database (July 2009

file). 3. Surveillance, Epidemiology, and End Results (SEER) Program SEER*Stat

Database: Incidence—SEER 17 Regs Limited-Use + Hurricane Katrina Impacted Louisiana Cases, Nov 2007 Sub (1973–2005 varying)—Linked To County Attributes—Total U.S., 1969–2005 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Cancer Statistics Branch, released April 2008, based on the November 2007 submission. Website: www.seer.cancer.gov.

4.1.4.3 Calculation of Base Incidence Rates Canadian Prostate Cancer statistics are available from Cancer Surveillance Online and the Canadian Cancer Registry. We used the Canadian Cancer Registry for the raw observed incidence rates. For staging to carve out stage A (T1a/T1b) we have used U.S. statistics from SEER Cancer Statistics Review since no Canadian data were available.

Raw Observed Incidence Rates The raw observed incidence rates for prostate cancer are published by Statistics Canada, Canadian Cancer Registry (CCR) Database (July 2009 file) and Demography Division (population estimates) (CANSIM table 103-0550).

Age and Sex Distribution of Prostate Cancer (All Stages) Distribution of Prostate Cancers Combined by Age Group, Canada, 2006

Age Group

2006 New Cases Raw incidence Rates (per 1,000)

Male Female Total Male Female Total 0-34 0 0 0 0 0 0 35-39 6 0 6 0.005 0 0.003 40-44 66 0 66 0.049 0 0.024 45-49 398 0 398 0.298 0 0.149 50-54 1320 0 1320 1.126 0 0.557 55-59 2803 0 2803 2.720 0 1.344 60-64 3900 0 3900 5.006 0 2.461 65-69 4454 0 4454 7.525 0 3.624 70-74 3759 0 3759 7.645 0 3.591 75-79 2819 0 2819 7.244 0 3.200 80-84 1767 0 1767 7.050 0 2.757 85+ 1183 0 1183 7.209 0 2.255

Total 22475 0 22475 1.390 0 0.689

Source: CANSIM table 103-0550, Canadian Cancer Registry (CCR) Database (July 2009 file).

http://www.seer.cancer.gov/�


39

We interpolated the incidence rates after all the adjustments. See section 4.1.5.10.

Definition Adjustment To derive stage A (T1a or T1b) prostate cancer we have to carve out these two stages from total prostate cancer incidence rates. Since there are no Canadian statistics of cancer types by stage (until 2008, provinces and territories were not required to report Canadian cancer data by stage and there is not much actual Canadian incidence experience by stage), we have used U.S. statistics by stage from SEER. We have analyzed numbers by stage.

The definition adjustment factor we have taken is the sum of the proportion of stage T1a and T1b. Therefore, it means that early stage prostate cancer accounts for 2.98% of all prostate cancer.

Source: U.S. SEER database frequency

An analysis of the SEER stage T1a/T1b data by age showed that the proportion of early stage prostate cancers to all prostate cancers varies materially. We decided to apply the proportions by age rather than use one adjustment over all ages.

Stage Count Stage DistT0 25 0.03%Tis 1 0.00%T1a 1 502 1.87%T1b 887 1.11%T1c 25 980 32.41%T1NOS 298 0.37%T2a 5 444 6.79%T2b 1 708 2.13%T2c 17 170 21.42%T2NOS 19 700 24.57%T3a 3 855 4.81%T3b 1 686 2.10%T3NOS 648 0.81%T4 1 259 1.57%Total 80 163 100%

Prostate Cancer Stage Distribution


40

Stage A (T1a/T1b) % by ages

Ages Proportion All ages 2.98%

0–34 0.00% 35–39 0.81% 40–44 0.81% 45–49 0.76% 50–54 0.67% 55–59 1.17% 60–64 1.61% 65–69 2.38% 70–74 3.16% 75–79 5.04% 80–84 7.30% 85+ 11.03%

Source: U.S. SEER database frequency 4.1.4.4 Trends There has been an overall upward trend in the incidence rate of prostate cancer since 1980, which is likely due to increased early detection. The SEER data provide information on the trend. We have also looked at Canadian incidence rates from year to year to determine the trend.

Based on SEER data, in the beginning of the 1980s there was a small increase. Between 1988 and 1993, the increase in the trend was significant because of the introduction of the PSA (Prostate Specific Antigen) test in 1988. After 1995, the trend decreased on older and increased on younger (early detection). Since 2000, the trend is downward for all ages.

We have analyzed the trend by age with Canadian and U.S. data. The trend is increasing for ages below 70 and decreasing for age 70 and over. This can be explain by earlier detection for younger ages and doctors not recommending the PSA test if patients have a life expectancy of less than 10 years for older ages.

We used the average actual annual trend from 1992 through to 2006 which we applied to the 2006 raw incidence rates to project them to 2008. These trends were determined by averaging the raw incidence rates from 1992 to 2006. Then we have projected the 2006 rates for two years by using these trends to come up with the 2008 Base Rates. Since this approach would trend the rates to July 1, 2008, we have backed out six months of trend to take the data back to January 1, 2008, by taking the square root of the 2008 annual trend factor. The following table is the result:


41

Source: Statistics Canada, Canadian Cancer Registry (CCR) Database (July 2009 file) and Demography Division.

4.1.4.5 First-ever Adjustment The Early Stage Prostate Cancer incidence rates were derived from the same data source as the Life-Threatening Cancer incidence rates. Therefore, the adjustment for Life-Threatening Cancer was used. See section 4.1.1.5.

4.1.4.6 Sudden Death Adjustment Since this is not life-threatening cancer, there is no adjustment for sudden death.


4.1.4.8 Prevalence Adjustment We derived early stage prostate cancer incidence rates from the same data source as the life-threatening cancer incidence rates. Therefore, the adjustment for life-threatening cancer was used. See section 4.1.1.8.

4.1.4.9 Thirty-day Survival Adjustment We used the standard approach described in section 3.8.

4.1.4.10 Interpolation/Fitting We interpolated the final adjusted incidence rates over the full age range using a three-point Lagrange interpolation formula. We did it after all adjustments because some of these adjustments were by age group, which creates scale rates.

The fitted rates were smoothed. The two adjustments were:

1. We zeroed the incidence rates between ages 12 and 35 since there was no cancer diagnoses at these ages; and

Year 0-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Total

1993 N/A 145.5% 128.8% 133.5% 144.0% 121.7% 122.6% 114.4% 101.0% 99.3% 99.6% 114.3%1994 N/A 81.3% 131.8% 124.7% 99.3% 100.9% 93.7% 90.8% 88.8% 86.2% 85.2% 93.8%1995 N/A 146.2% 97.3% 81.4% 87.7% 85.8% 85.8% 82.4% 84.4% 86.1% 100.1% 86.7%1996 N/A N/A 102.8% 111.7% 109.6% 102.6% 99.4% 100.2% 94.5% 95.5% 89.4% 100.1%1997 N/A N/A 113.4% 112.6% 107.6% 107.0% 106.1% 103.7% 105.8% 103.2% 99.5% 106.6%1998 N/A 118.8% 113.4% 110.2% 112.0% 103.3% 101.9% 94.0% 91.3% 97.8% 102.9% 101.0%1999 N/A 126.3% 109.0% 108.0% 104.3% 106.8% 105.8% 104.8% 106.5% 95.4% 96.3% 105.8%2000 N/A 112.5% 119.7% 107.6% 107.1% 106.5% 102.8% 105.4% 104.4% 98.0% 104.4% 106.1%2001 N/A 163.0% 116.5% 122.7% 120.9% 109.5% 113.2% 101.4% 99.4% 102.1% 96.3% 108.8%2002 N/A 84.1% 110.0% 98.7% 91.0% 97.3% 90.2% 94.2% 89.3% 92.7% 93.7% 94.6%2003 N/A 110.8% 103.3% 106.3% 97.2% 98.6% 98.1% 95.3% 95.8% 96.2% 95.2% 99.3%2004 N/A 126.8% 108.8% 100.2% 112.1% 100.9% 103.8% 100.3% 98.7% 99.2% 96.7% 104.0%2005 N/A 92.3% 100.7% 100.9% 99.2% 101.4% 101.3% 97.9% 96.6% 97.9% 98.1% 101.2%2006 N/A 102.1% 109.2% 112.6% 106.3% 107.0% 103.9% 103.6% 101.2% 96.6% 91.1% 105.5%2007 N/A 111.3% 111.4% 108.7% 106.3% 103.2% 101.7% 98.9% 96.8% 96.0% 96.2% 101.8%2008 N/A 105.5% 105.5% 104.2% 103.1% 101.6% 100.8% 99.4% 98.4% 98.0% 98.1% 100.9%

2008/2006 N/A 117.4% 117.5% 113.3% 109.5% 104.9% 102.5% 98.3% 95.2% 94.1% 94.3% 102.6%

Canadian Male Annual Trend for Prostate Cancer


42

2. We capped incidence rates above age 87 at the age 87 incidence rate. We did this because we did not want the maximum incidence rate to be far from the maximum adjusted incidence rate and there was no data breakdown by age after age 85, so we do not know how the incidence moves upward or downward.


Central Age 60 70 80Base Rate 4.0086 7.8849 7.1028 N/A N/A N/A

Adjustments:Definition Adjustment -98.4% -96.8% -92.7% N/A N/A N/A2008 Trend 4.9% -1.7% -5.9% N/A N/A N/AFirst Ever Incidence -6.8% -12.5% -17.7% N/A N/A N/ASudden Death 0.0% 0.0% 0.0% N/A N/A N/AOverlap 0.0% 0.0% 0.0% N/A N/A N/APrevalence 6.1% 14.4% 18.7% N/A N/A N/A30-Day Survival -0.1% -0.2% -0.5% N/A N/A N/A

Derived Incidence Rate 0.0673 0.2499 0.4915 N/A N/A N/A

Note: The Derived Incidence Rate = (Raw Observed Rate)x (1 + Definition Adjustment)x (1 + 2008 Trend Adjustment)x (1 + First Ever Incidence Adjustment)/ (1 - Sudden Death Adjustment)x (1 + Overlap Adjustment)/ (1 - Prevalence Adjustment)x (1 + 30-Day Survival Adjustment)

Summary of Calculation of Incidence Rates per 1,000: Early Stage Prostate CancerMales Females


43

A comparison of the raw and derived incidence rates per 1,000 for ages 15—100 is graphed below.

Other Considerations Early Detection:

There are some unique characteristics to prostate cancer that could increase the anti-selection risk in the pricing CI product. Prostate cancer is one cancer that can jeopardize the pricing of CI product. Many prostate cancers are undetected. Some tests are now available at younger ages, and more doctors are asking their patients to have these tests. The prostate cancer raw rates in these tables do not reflect undetected/latent (occult) prostate cancer where prevalence rates of these occult cancers have been estimated to be as high as 34% in men aged 40–49 (see here). Since early stage prostate cancer is excluded from the life-threatening cancer definition but paid as partial benefit, it should be monitored against emerging experience.

Canada versus U.S.:

We used raw incidences rates of the Canadian population but the trend adjustment comes from U.S. statistics. The Canadian universal health care system and access of treatments and screening in Canada are quite different to those in the U.S. We have, nonetheless, assumed U.S. numbers to be representative of Canada.

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

15 20 25 30 35 40 45 50 55 60 65 70 75 80 85

Rate

per

1,0

00

Age Last

Early Stage Prostate Cancer Incidence Rates Initial vs Final Basis

Initial Male Final Male

http://www.cancer.gov/cancertopics/pdq/prevention/prostate/healthprofessional/page2�


44

4.1.5 Ductal Carcinoma in Situ 4.1.5.1 Definition Ductal carcinoma in situ (DCIS) is defined as ductal carcinoma of the breast as confirmed by biopsy. The diagnosis must be confirmed by a pathologist.






A claimant must report this medical information to the company within six months of the date of diagnosis. If the claimant does not provide this information, the company has the right to deny any claim for ductal carcinoma in situ or any critical illness caused by any cancer or its treatment.

4.1.5.2 Data Sources 1. Canadian Cancer Statistics, 2007. 2. SEER Cancer Statistics Review, National Cancer Institute:

Area 17, all ethnic races, 2000–2004. Table IV – 5 Female Breast Cancer Invasive. Table IV – 7 Female Breast Cancer (In Situ).

4.1.5.3 Calculation of Base Incidence Rates There are no detailed statistics for the incidence of DCIS in Canada as confirmed by the Canadian Cancer Registry confirmed that these statistics are not available. There are, however, detailed U.S. statistics by age and race for all in-situ breast cancers. Based on the U.S. data, we used the following general approach to develop the incidence rates for DCIS for Canada:

1. We examined the detailed U.S. SEER incidence rates for in-situ female breast cancers and total invasive breast cancers.

2. We determined the proportion of the in-situ cancers that would be classified as DCIS under the Canadian critical illness definition of DCIS.

3. Using SEER data, we created ratios of DCIS incidence to total invasive breast cancer incidence.

4. We applied these ratios to the Canadian cancer statistics to develop Canadian DCIS incidence rates.

5. We applied trend factors and other adjustments.


45

This method assumes the incidence of DCIS in Canada as a proportion of all invasive breast cancers is the same as in the U.S. It also assumes that the proportion of DCIS among all in-situ breast cancers is the same in Canada and the U.S. Given the materiality of the DCIS rates as compared to total critical illness incidence, we are comfortable making these assumptions. The process of developing rates using the method described is more fully explained below.

Raw Observed Incidence Rates We based the raw observed incidence rates on SEER Female Breast Cancer Table IV-5 (invasive) and IV-7 (in situ, all races). The age-specific breast cancer rates, including the ratio of in-situ to invasive cancers, are as follows:

The ratios peak in the 40s presumably because extensive mammogram screening programs make early detection of DCIS more likely.

Examination of In-situ Breast Cancers Included in DCIS Definition Not all in-situ breast cancers would meet the definition of DCIS which would result in a claim payment. The SEER data provide a detailed breakdown of the various types of in-situ breast cancers which can be used to estimate what percentage of the in-situ data should be used for the purposes of developing the DCIS rates.

Female Breast Cancer

Source: SEER 17 Area: Age Specific Rates: All races, 2000 – 2004

Incidence rates per 100,000

Age In situ Invasive Ratio: In situ to Invasive

15-19 0.2 20-24 0.2 1.4 14% 25-29 1 7.7 9%

30 – 34 2 25.5 9% 35-39 10 59.5 17%

40 – 44 34 117.4 29% 45 – 49 54 186.7 29% 50 – 54 67 237.3 28% 55 – 59 80.3 307.9 26% 60 – 64 86.7 363.7 24% 65 - 69 97 410.9 24% 70 – 74 92.4 432.3 21% 75 – 79 86.5 464.8 19% 80 – 84 65 435.2 15%

85+ 33.9 355.3 10%


46

Percentage Distribution and counts by histology among histologically confirmed cases (2001–2004):

In general all of the histologies classified as adeno-carcinomas would be considered DCIS, except for lobular carcinoma in situ, which represents 11% of the total. The “other” in-situ histologies are also not classified as DCIS and they represent 20% of the total. Therefore, DCIS as defined for critical illness represents 69% of all in-situ histologies. We used 70% as the adjustment factor to apply to the ratio of in-situ to invasive cancers. There is not sufficient information to have factors that vary by age.

4.1.5.4 Trends The SEER data also provide some information on the trend of DCIS over time. No information specific to Canadian trends in DCIS were found so the SEER information was assumed to be representative of Canadian trends. From 1995 to 2004, the observed trend (average percentage change) was 2.6%. This compares with a trend of -1.0% for invasive breast cancer over the same time period. Due to the significant difference in trends, the rate of 2.6% is used to project the data forward from the time period 2000 to 2004, to the base year of 2008 (5.5 years in total).

4.1.5.5 First-ever Adjustment The DCIS incidence rates were derived from the same data source as the life-threatening cancer incidence rates. In addition, since DCIS is more common at older ages, it is possible for an individual to contract another cancer, recover and then get DCIS. In fact, the chemotherapy and radiation treatment for some cancers (e.g., Hodgkin’s lymphoma)

ALL RACES

Count Percent

Adenocarcinoma in situ 16 0%

Cribiform Carcinoma in situ 3674 8%

Ductal Carcinoma in situ 18368 40%

Lobular Carcinoma in situ 5244 11%

Comedo-carcinoma in situ 4518 10%

IntraDuctal and lobular Carcinoma in situ 1867 4%

IntraDuctal micropapillary Carcinoma in situ 1613 3%

Other Adenocarcinoma 1520 3%

All Adenocarcinoma 36820 80% Other in situ histologies 9305 20% TOTAL 46125 100%


47

may increase the likelihood of the treated individual getting DCIS. Due to data limitations, the adjustment for life-threatening cancer was used. See section 4.1.1.5.

4.1.5.6 Sudden Death Adjustment As with life-threatening cancer, no adjustment was made for sudden death. See section 4.1.1.6.


4.1.5.8 Prevalence Adjustment DCIS incidence rates were derived using the same data source as the life-threatening cancer incidence rates. Therefore, the adjustment for life-threatening cancer was used. See section 4.1.1.8.

4.1.5.9 Thirty-day Survival Adjustment As with life-threatening cancer, the mortality for normal lives was used. See section 4.1.1.9.


Central Age 35 55 75 35 55 75Base Rate 0.0000 0.0000 0.0000 0.0463 0.6787 0.7123

Adjustments:Definition Adjustment 0.0% 0.0% 0.0% -30.0% -30.0% -30.0%2008 Trend 0.0% 0.0% 0.0% 15.2% 15.2% 15.2%First Ever Incidence 0.0% 0.0% 0.0% -1.9% -7.0% -13.2%Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Prevalence 0.0% 0.0% 0.0% 1.1% 5.1% 11.1%30-Day Survival 0.0% 0.0% 0.0% 0.0% 0.0% -0.2%


Check 0.0000 0.0000 0.0000 0.0370 0.5365 0.5592

Note: The Derived Incidence Rate = (Raw Observed Rate)x (1 + Definition Adjustment)x (1 + 2008 Trend Adjustment)x (1 + First Ever Incidence Adjustment)/ (1 - Sudden Death Adjustment)x (1 + Overlap Adjustment)/ (1 - Prevalence Adjustment)x (1 + 30-Day Survival Adjustment)

Summary of Calculation of Incidence Rates per 1,000: Ductal Carcinoma In SituMales Females


48

4.2 Heart Heart (cardiovascular) conditions accounted for more than 20% of critical illness claims, led by heart attack (acute myocardial infarction) claims, second only to cancer in magnitude. Accordingly, cardiovascular condition incidence rates are a vital component of pricing critical illness coverage.

Dramatic changes in the medical landscape for cardiovascular conditions and medical procedures in Canada since 2000 make it challenging to derive incidence rates. There has been a significant decrease in age-adjusted incidence of both heart attack and stroke, which may be the result of changes in lifestyle, treatment, medication, and better education. The clinical definition of heart attack has changed. The use of certain covered procedures shifted due to changes in risk factors and popularity or efficacy of those procedures relative to other procedures. As an example, there has generally been a decrease in the use of coronary artery bypass surgery (CABG) and an increase in the use coronary angioplasty, although rates for the latter recently decreased for the first time in many years.

We have used the best available data to provide a basis that should be appropriate as of January 2008. Given the rapid pace of change experienced in the last decade it will be important to monitor claims experience in these categories carefully.

In order to determine the overlap adjustment for cardiovascular conditions, the order of precedence is assumed to be heart attack, stroke, CABG and coronary angioplasty respectively, and that overlap adjustments would be made accordingly. The order is consistent with the number of claims received according to the Munich Re 2010 survey. No adjustment is required for heart attack. For stroke, an adjustment for heart attacks will be required, while adjustments for heart attack and stroke will be required for CABG.

4.2.1 Heart Attack (Acute Myocardial Infarction) 4.2.1.1 Definition The medical definition of a heart attack has changed over the past 10 years, and critical illness policy definitions have been updated accordingly. However, these updates are typically applied to new policies, and existing policies may or may not adopt the new definitions. Accordingly, the definition chosen for this exercise is representative of those used in fully-underwritten non-cancellable Canadian critical illness policies in January 2008:

Heart attack (acute myocardial infarction or AMI) shall mean the diagnosis of the death of a portion of heart muscle as a result of inadequate blood supply as evidenced by:

a) new electrocardiographic (ECG) changes, indicative of a myocardial infarction; and b) elevation of cardiac biochemical markers to levels considered diagnostic for acute infarction.

Heart attack does not include:

a) an incidental finding of ECG changes suggesting a prior myocardial infarction, in the absence of a corroborating event; or b) elevation of cardiac markers to coronary angioplasty unless there are diagnostic changes of new Q wave infarction on the ECG.


49

A 30-day survival period is usually required.

The CLHIA Benchmark Definition for Heart Attack, which is expected to apply to most new fully-underwritten non-cancellable policies by 2009, is materially different from the above definition. The benchmark definition requires a rise and fall of biochemical markers to diagnostic levels, plus any one of ECG changes, heart attack symptoms, or development of new Q waves as a result of an intra-arterial cardiac procedure. As discussed above, this definition will not be consistent with the majority of the policies considered in our experience study, and accordingly it has not been adopted for the development of the incidence basis.

4.2.1.2 Data Sources 1. Canadian Institute for Health Information, special request (Ontario Information

for period from 2004 to 2006). 2. Surveillance Division, Centre for Chronic Disease Prevention and Control.

Chronic Disease Infobase. 3. Health Canada. Tobacco Control Programme Canadian Tobacco Use Monitoring

Survey (CTUMS), Smoking Prevalence 1999–2007. 4. Statistics Canada. Annual Demographic Estimates: Canada, Provinces, and

Territories, 2007 Revised. Catalogue no. 91-215 X, CD Data file. 5. Manuel et al. “How Many People have had a Myocardial Infarction? Prevalence

Estimated Using Historical Hospital Data”. BMC Public Health 7 (2007): 174. 6. Canadian Institute for Health Information. Health Care in Canada 2007. 7. Institute for Clinical and Evaluative Studies (ICES). Website: www.ices.on.ca. 8. Tu et al. “Outcomes of Acute Myocardial Infarction in Canada”. Canadian

Journal of Cardiology, Vol. 19 No. 8 (July 2003). 9. Alter et al. “Socioeconomic Status and Mortality after Acute Myocardial

Infarction”. Annals of Internal Medicine, Vol. 144 No. 2 (2006): 82–93. 10. Koek et al. “Acute myocardial infarction incidence and hospital mortality:

routinely collected national data versus linkage of national registers”. European Journal of Epidemiology, Vol. 22 No. 11 (2007): 755–762.

11. Dinani et al. A Critical Review. Staple Inn Actuarial Society, 2000. 12. Heart and Stroke Foundation of Canada. The Growing Burden of Heart Disease

and Stroke in Canada 2003. 13. Robjohns et al. Exploring the Critical Path. Staple Inn Actuarial Society, 2006. 14. Statistics Canada. Population Projections for Canada, Provinces and Territories,

2005–2031. Catalogue no. 91-520-XIE.

4.2.1.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates Given continuing changes in the diagnosis and definition of heart attack, it was deemed important to use the most recent available data to develop the incidence basis. In Canada the Canadian Institute for Health Information (CIHI) compiles statistics for each province on hospital discharges, but a lag exists between the date the information is gathered and the date the information is released. We requested a special tabulation of all discharges

http://www.ices.on.ca/�


50

for ICD codes I-21 (first-time acute myocardial infarction (AMI)) and I-22 (repeat AMI) from the CIHI for data from 2004 through 2006. To reduce the cost we confined the request to Ontario, the most populous province in Canada, and then made subsequent adjustments to provide Canadian incidence rates.

Discussions with the CIHI confirmed that I-21 refers to first instance of AMI during that admission to the hospital—which is not consistent with a “first-ever” definition. Accordingly, the information for code I-22, which refers to a subsequent infarction during that admission, was ignored in our analysis. Data was provided for all AMI cases, those where AMI was the most responsible diagnosis, and split these into acute care admissions, emergency room admissions, and day surgery admissions. The data for acute care admissions, where AMI was the most responsible diagnosis, was chosen as best representing the definition shown above and this basis is consistent with AMI statistics reported in Canada in such publications as The Growing Burden of Heart Disease and Stroke in Canada 2003. The following chart provides the acute care admission counts for Ontario from 2004 to 2006, where AMI was the most responsible diagnosis:

Male Female 2004 2005 2006 2004 2005 2006

AGE

0-19 5 4 6 2 1 - 20-24 12 9 11 4 4 3 25-29 27 35 27 3 5 1 30-34 84 91 59 18 16 26 35-39 245 218 224 56 52 45 40-44 664 627 547 148 145 136 45-49 1,142 1,071 966 264 277 271 50-54 1,541 1,418 1,392 384 335 361 55-59 1,749 1,784 1,588 560 530 513 60-64 1,657 1,574 1,609 644 630 582 65-69 1,684 1,529 1,382 830 816 717 70-74 1,846 1,684 1,530 1,192 1,076 946 75-79 1,926 1,721 1,630 1,503 1,406 1,285 80-84 1,623 1,559 1,557 1,679 1,677 1,562 85-89 968 917 866 1,214 1,238 1,107 90+ 397 389 375 829 776 744

We chose to use the 2006 data in order to estimate incidence rates. This reflects the fact that the number of heart attacks in that year is significant for most quinquennial age bands, and the desire to capture the most recent information.

The number of discharges for condition I-21 was then divided by the estimated population of Ontario in 2006 to provide crude rates. We used the Statistics Canada population estimate rather than the census figure, since the census figure does not include a significant portion of the population which is not enumerated.


51

Age and Sex Distribution of Heart Attack Distribution by Age Group and Gender, Ontario, 2006

Age Group

2006 New Case Estimates (data source 1)

2006 Population Estimates (data source 4)

Raw incidence Rates

(in 1,000s) (per 1,000) Male Female Total Male Female Total Male Female Total

0–19 6 0 6 1,595 1,523 3,118 0.0038 0.0000 0.0019 20–24 11 3 14 444 430 874 0.0248 0.0070 0.0160 25–29 27 1 28 424 425 849 0.0637 0.0024 0.0330 30–34 59 26 85 446 446 892 0.1323 0.0583 0.0953 35–39 224 45 269 480 483 963 0.4667 0.0932 0.2793 40–44 547 136 683 546 541 1,087 1.0018 0.2514 0.6283 45–49 966 271 1,237 511 514 1,025 1.8904 0.5272 1.2068 50–54 1,392 361 1,753 434 448 882 3.2074 0.8058 1.9875 55–59 1,588 513 2,101 383 397 780 4.1462 1.2922 2.6936 60–64 1,609 582 2,191 287 299 586 5.6063 1.9465 3.7389 65–69 1,382 717 2,099 223 243 466 6.1973 2.9506 4.5043 70–74 1,530 946 2,476 188 213 401 8.1383 4.4413 6.1746 75–79 1,630 1,285 2,915 151 189 340 10.7947 6.7989 8.5735 80–84 1,557 1,562 3,119 98 151 249 15.8878 10.3444 12.5261 85–89 866 1,107 1,973 44 84 128 19.6818 13.1786 15.4141 90+ 375 744 1,119 17 47 64 22.0588 15.8298 17.4844

Total 13,769 8,299 22,068 6,271 6,433 12,704 2.1957 1.2901 1.7371

We compared information from the Chronic Disease Infobase (data source 2) with the age-standardized incidence rates for Ontario to those of Canada. Ontario rates have traditionally been higher, but this difference has reduced since 2000, and the male rates were about 1% higher in Ontario than across Canada in 2005. In 2005 female rates were approximately 8% higher in Ontario than in Canada, compared to a difference of 16% in 2001. Between 2000 to 2005, the average annual change in the ratio was approximately a 1% decrease for males and no change for females. Combining these factors, the Ontario to Canada adjustment factors are 100% for males and 92.6% for females (Canadian female rates are assumed to be 92.6% of Ontario rates).

We then recalculated the incidence rates using the Ontario-to-Canada adjustment and placed them into age bands consistent with those employed for the other conditions in this paper.


52

Age Group

Raw incidence Rates (per 1,000)

Male Female Total 0–19 0.0038 0.0000 0.0019 20–29 0.0438 0.0043 0.0242 30–39 0.3056 0.0708 0.1880 40–49 1.4314 0.3572 0.8948 50–59 3.6475 0.9577 2.2799 60–69 5.8647 2.2191 3.9865 70–79 9.3215 5.1387 7.0523 80–89 17.0634 10.5162 12.9822 90+ 22.0588 14.6572 16.6233

Total 2.1957 1.1945 1.6887

The raw incidence rates are defined as the number of new heart attacks per 1,000 persons in the population. These raw incidence rates have been interpolated/extrapolated over the full age range using a three-point Lagrange interpolation formula.

Adjustment for Heart Attacks Excluded from the Definition We would expect the definition of heart attack used in critical illness policies to exclude some heart attacks which are diagnosed and included in the discharge statistics above. However, the clinical definition and diagnostic methods associated with heart attacks have changed dramatically since 2000. Early contract definitions are no longer in line with current accepted medical practice for treatment of heart attacks, and may require diagnostic procedures which are no longer performed. The original intent was to identify and pay claims for legitimate heart attacks, which would not include incidental findings without corroborating symptoms at the date of the attack. Accordingly many critical illness providers have adjusted their interpretation beyond contract terms, to ensure they are capturing legitimate claims that meet the spirit of the definition. The introduction of troponin as the gold standard for measuring heart muscle damage was expected to lead to a significant increase in heart attack rates. Based on the statistics to date, this increase does not appear to have occurred, although there may be many reasons for this result. This leads to uncertainty as to whether an adjustment to remove heart attacks, which may not be covered by critical illness policies, is warranted at this point. To be conservative, we have chosen not to adjust the underlying data.

4.2.1.4 Trends The data require adjustment for trend from October 2006 (the midpoint of the study period for the base data) to January 2008. Information from the Chronic Disease Infobase (data source 2) suggests age-adjusted incidence has decreased sharply over the past five years, which would normally necessitate a trend adjustment to reduce the incidence rate.


53

Year Ontario -

Male Canada -

Male Ontario - Female

Canada - Female

1994 277.6 268.2 124.5 113.9 1995 280.7 272.5 127.7 115.2 1996 293.0 277.4 133.1 117.7 1997 287.6 268.1 128.4 114.1 1998 285.9 264.1 123.5 108.9 1999 284.2 263.6 128.4 111.9 2000 284.7 266.6 126.5 114.1 2001 283.9 259.8 130.2 112.2 2002 277.3 254.0 125.9 109.3 2003 254.4 241.0 116.4 105.4 2004 236.5 224.5 109.9 99.6 2005 215.5 213.9 103.0 95.3

In some jurisdictions events such as smoking bans have led to one-time immediate decreases in AMI incidents. Ontario has led an aggressive campaign against smoking over the past few years, which resulted in a significant reduction in smoking prevalence in 2005 and 2006, but early indications show Ontario smoking prevalence increased in 2007. There has also been progress in education regarding cardiovascular health, and increased monitoring for risk factors—this may have the effect of deferring heart attacks rather than eliminating them altogether. We are also faced with the increase in diabetes and obesity, which will tend to increase AMI incidence rates in the long term. Finally, changes in the definition of heart attack may have some role to play in the reduction we have seen in incidence rates. Based on all of these factors, we have decided not to adjust the incidence rates for the mathematically-indicated trend adjustment.

The sharp reduction in incidence discussed above led to debate among the members of the subcommittee regarding the accuracy of the data and their application in the context of developing a population incidence table for critical illness. The data come from the best Canadian source, and we have verified that the base information is consistent with prior reported periods. It will be important to monitor this closely over the next few years to understand the forces behind these reductions, and their relationship to critical illness incidence rates.

Age Standardized Incidence Rate for Acute Myocardial Infarction

050

100150200250300350

1992 1994 1996 1998 2000 2002 2004 2006Rate

per

100

,000

pop

ulat

ion

Ontario - MaleCanada - MaleOntario - FemaleCanada - Female


54

4.2.1.5 First-ever Adjustment The Institute for Clinical Evaluative Sciences (ICES) publishes estimated incident and prevalent cases of heart attacks in Ontario. The incident cases are based on hospital registry records, excluding cases where a prior heart attack was recorded after 1988 (the first year for hospital records in the ICES study). The number of cases is then adjusted to account for incomplete data prior to the beginning of the study period.

A review of the incident cases versus the base incidence counts obtained from CIHI implied a very high percentage of cases were incident cases, particularly when compared to other studies such as the Staple Inn report. However, given the decreases in overall incidence described above, and to be conservative, we chose to use the ICES data to develop the first-ever adjustment. Lagrange interpolation was performed to provide first-ever ratios by age.

ICES Incident Cases (data source 7)

CIHI Ontario Cases (data source 1) First-ever Ratio

Age Male Female Male Female Male Female 20–49 1,499 416 1,834 482 82% 86% 50–64 3,779 1,367 4,589 1,456 82% 94% 65–79 3,802 2,864 4,542 2,948 84% 97% 80+ 2,397 3,407 2,798 3,413 86% 100%

We have noted above that ICD code I-21 does not provide information regarding first-ever AMI, but rather first incident during a hospitalization. This appears to be unique to Canadian reporting. In other jurisdictions, I-21 is regarded as a first-ever measure. In the paper Exploring the Critical Path (data source 13), as an example, the ratio of (I-21)/(I-21 + I-22) is used as an estimate of first-ever proportion, which yields a figure of 84%. This is in contrast to our Canadian figure of 99.4% for the same ratio.

4.2.1.6 Sudden Death Adjustment There is no sudden-death adjustment required, since we have used only acute-care discharge data for those who were admitted to hospital.

4.2.1.7 Overlap Adjustment We determined that for the “cardiac conditions”, the order of precedence would be assumed to be heart attack, stroke, CABG, and angioplasty respectively, and that overlap adjustments would be made accordingly. No adjustment is required for heart attack. For stroke, an adjustment for heart attacks will be required, while adjustments for heart attack and stroke will be required for CABG.

4.2.1.8 Prevalence Adjustment The population needs to be adjusted to remove all individuals who have previously been diagnosed with heart attack.

The ICES publishes estimated prevalence rates for the Ontario population. The 2006/7 estimates were as follows:


55

ICES Heart Attack Prevalence(data source 7)

Age Male Female 20–49 0.39% 0.10% 50–64 4.22% 1.08% 65–79 10.63% 4.38% 80+ 15.84% 9.47%

We used a Lagrange interpolation to produce rates by age. A weighted average comparison of incidence rates from 1994 to 2004 from the Chronic Disease Infobase (data source 2) was then used to estimate the required adjustment from Ontario data to Canadian data. The rates increase linearly from 50% in 1994 to 100% in 2004.

4.2.1.9 Thirty-day Survival Adjustment We estimated 30-day survival rates using the paper “Outcomes of Acute Myocardial Infarction in Canada” (data source 8), which uses CIHI data and has information by age group. This data covers the period from 1997/1998 to 1999/2000 (using CIHI’s fiscal year) and is based on the chart shown in figure 2 of that paper.

Mortality Rates by Age-Band and Sex

Age Band Female Male 20–24 0.0% 2.2% 25–29 0.0% 1.7% 30–34 1.3% 1.1% 35–39 3.8% 1.1% 40–44 2.5% 1.2% 45–49 3.8% 2.1% 50–54 4.3% 2.6% 55–59 5.4% 3.7% 60–64 7.2% 5.3% 65-69 10.0% 8.7% 70–74 14.5% 11.9% 75–79 17.6% 16.5% 80–84 24.4% 23.0% 85–89 29.6% 29.7% 90+ 36.1% 38.9%

We trended the data to 2006 using aggregate statistics based on CIHI data, from the Health Care in Canada report (data source 6). They noted that over a five-year period, from 1999/2000 to 2004/2005, the age-standardized 30-day mortality rate fell from 13.4% to 11.1%. This suggests annual improvement of approximately 3.7%. This improvement factor was then applied to trend the survival rates from 1998/1999 to 2006/2007, which results in a 26% improvement in the age-specific survival rates from 1998/1999. Since no further trend was assumed for the incidence rates from 2006 to 2008, none is assumed for survival.

We used a Lagrange interpolation to adjust this for our incidence basis.


56


Summary of Calculation of Incidence Rates per 1,000: Heart Attack Males Females Central Age 35 55 75 35 55 75 Raw Observed Rate 0.34 3.76 9.62 0.09 1.09 5.78

Adjustments: Ontario to Canada Adjustment

0.00% 0.00% 0.00% -7.40% -7.40% -7.40%

Heart Attacks Excl. from Definition

0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

2008 Trend 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% First-ever Incidence -23.8% -18.2% -15.9% -13.7% -6.2% -2.1%

Sudden Death 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% Overlap 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% Prevalence 0.2% 3.5% 11.8% 0.0% 0.8% 4.9% 30-Day Survival -0.8% -2.4% -10.7% -2.0% -3.6% -12.2%



57


4.2.2 Coronary Artery Bypass Graft (CABG) 4.2.2.1 Definition CLHIA Benchmark Definition Coronary artery bypass surgery is defined as the undergoing of heart surgery to correct narrowing or blockage of one or more coronary arteries with bypass graft(s), excluding any non-surgical or trans-catheter techniques such as balloon angioplasty or laser relief of an obstruction. The surgery must be determined to be medically necessary by a specialist.

4.2.2.2 Data Sources 1. ICES (Institute for Clinical and Evaluative Studies). Access to Health Services in

Ontario 2nd Edition. 2. Heart and Stroke Foundation of Canada. The Growing Burden of Heart Disease

and Stroke in Canada 2003. Table 2-23, page 42. 3. Statistics Canada. Annual Demographic Statistics, 2005. Catalogue no. 91-213-

62. 4. The Staple Inn Actuarial Society. A Critical Review: Report of the Critical Illness

Healthcare Study Group. 5. Canadian Institute for Health Information. CIHI Health Indicator Reports CABG. 6. CIHI. Health Indicators 2009.

0,0

5,0

10,0

15,0

20,0

25,0

30,0 R

ate

per 1

,000

Age Last

Heart Attack Incidence Rates Raw vs Derived

Raw Male Raw Female Derived Male Derived Female


58

7. William A. Ghali for the Canadian Cardiovascular Outcomes Research Team. “Outcomes of Coronary Artery Bypass Graft Surgery in Canada 1992/1993 to 2000/2001”. Canadian Journal of Cardiology, Vol. 19 No. 7 (June 2003).

8. The Institute for Evaluative Sciences, Toronto, in collaboration with Informatics Committee of the Cardiac Care Network of Ontario. Report on Coronary Bypass Surgery in Ontario Fiscal Years 2002–2004. Tables 4(a), 4(b) and 4(c).

9. “Multicenter Experience in Revascularization of Very Elderly Patients”. American Heart Journal Vol. 148 No. 3 (2004): 486–492.

10. Revascularization and Survival Trends of Canadian Acute Myocardial Infarction (AMI) Patients, 1995/1996 to 2003/2004.

4.2.2.3 Calculation of Base Incidence Rates Base Data We used base data from ICES’ (Institute for Clinical and Evaluative Studies) Access to Health Services in Ontario 2nd Edition (Chapter 3, page 66, table 3.3c), which used year 2004/2005 results based on sex and four age groupings.

CABG Incidence per 100,000 Male Female

20–39 2 0 40–64 145 25 65–74 526 150 75+ 372 98

These were interpolated using a three-point Lagrange formula.

Ontario to Canada Adjustment We used the same data as for trend below (CIHI Health Indicator Reports). Both males and females produced a 98.2% adjustment.

4.2.2.4 Trends We obtained overall Canadian incidence rate from CIHI for years 1999 to 2005 (data source 4).

Canada Age-Standardized Incidence Rates (per 100,000) Male Female 1999 153.1 41.9 2000 154.6 43.0 2001 155.4 41.5 2002 151.5 40.4 2003 152.5 39.9 2004 145.7 37.0 2005 137.6 35.3

By using a logarithmic regression analysis, we determined that the average decrease in incidence was 1.76% for males and 1.93% for females. Applying this to the 2.75 year


59

period from 2005 to 2008 produces a reduction of 4.43% for males and 8.17% for females. As a check we used parallel Ontario CIHI data. In this case the average decrease was 1.51% for males and 1.96% for females.

Health Indicators 2009 (data source 6) shows a continued reduction, not sex-specific as shown below:

Rates of Cardiac Revascularization Procedures, Canada (Age-standardized rate per 100,000, excluding Québec, ages 20 and older)

1998–1999 91 1999–2000 93 2000–2001 93 2001–2002 93 2002–2003 91 2003–2004 90 2004–2005 86 2005–2006 84 2006–2007 77 2007–2008 75

The annual reduction rate is thus 2.21% since 1998–1999 and 3.52% since 2001–2002.

Thus the trend reductions are conservative.

4.2.2.5 First-ever Adjustment Outcomes of coronary artery bypass graft surgery in Canada 1992/1993 to 2000/2001 (data source 5) indicated the percentage of patients with a prior bypass graft was 9%. However, it gave no indication of incidence by age. We thus used U.S. data supplied by the article “Multicenter Experience in Revascularization of Very Elderly Patients” (data source 7). Again a three-point Lagrange formula provided rates from ages 35 to 85. The rate was assumed to remain level at older ages.

4.2.2.6 Sudden Death Adjustment We did not adjust for sudden death since all data related to hospital admission and medical procedures and should not be affected by sudden death.

4.2.2.7 Overlap Adjustment We used data from Revascularization and Survival Trends of Canadian Acute Myocardial Infarction (AMI) Patients, 1995/1996 to 2003/2004.

Approximate CABG Rates in 2003/2004 Following AMI

Rates Used Mid-age Male Female 20–44 7% 5% 38 45–64 12% 9% 58 65–79 13% 9.00% 72 80+ 3% 1.50% 83

4.2.2.8 Prevalence Adjustment We used the same prevalence adjustment as that used for AMI.


60

4.2.2.9 Thirty-day Survival Adjustment Our source data was from the Report on Coronary Bypass Surgery in Ontario Fiscal Years 2002–2004 (data source 6), which provided mortality rates in three bands: under 65; 65–74 and 75 and older. We assumed no mortality to age 30 and then used a three-point Lagrange formula to interpolate rates to age 100.


4.2.3 Coronary Angioplasty

4.2.3.1 Definition The definition which is representative of those used in fully-underwritten non-cancellable Canadian critical illness policies in January 2008 is as follows:

Coronary angioplasty is a medically necessary, non-surgical interventional procedure to unblock and widen a vessel to allow an uninterrupted flow of blood and oxygen to the heart. The procedure must be performed by a physician.

Summary of Calculation of Incidence Rates per 1000: Coronary Artery Bypass Graft (CABG)Males Females


Adjustments:Definition -1.8% -1.8% -1.8% -1.8% -1.8% -1.8%2008 Trend -4.4% -4.4% -4.4% -8.2% -8.2% -8.2%First Ever adjustment 0.0% -7.0% -9.0% 0.0% -7.0% -9.0%Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Overlap -6.3% -11.3% -10.3% -4.4% -8.4% -6.8%Prevalence 0.2% 3.5% 11.8% 0.0% 0.8% 4.9%30-day Survival -0.2% -0.5% -1.5% -0.2% -0.9% -2.7%


Check 0.02708 1.75730 4.54761 0.00000 0.26851 1.27331

Note: the derived Incidence Rate = (Raw Observed Rate)x (1 + Definition Adjustment)x (1+ 2008 Trend Adjustment)x (1 + First Ever Adjustment)x (1 - Sudden Death Adjustment)x (1 + Overlap Adjustment)x (1 - Prevalence Adjustment)x (1 + 30-Day Survival Adjustment)


61

A 30-day survival period is also representative.

4.2.3.2 Data Sources 1. ICES. Access to Health Services in Ontario 2nd edition. April 2005. 2. CIHI Health Indicators, 2006 and 2007. 3. Statistics Canada. Population Projections for Canada, Provinces and Territories,

2005–2031. Catalogue no. 91-520-XIE. 4. Anderson et al. “A Contemporary Overview of Percutaneous Coronary

Interventions”. Journal of the American College of Cardiology Vol. 39 No. 7 (2002).

5. Johansen et al. “Revascularization and Heart Attack Outcomes”. Statistics Canada Health Reports Vol. 13 No. 2 (January 2002).

6. Dinani et al. A Critical Review. Staple Inn Actuarial Society, 2000. 7. Canadian Cardiovascular Outcomes Research Team. Canadian Cardiovascular

Atlas.

4.2.3.3 Calculation of Incidence Rates Raw Observed Incidence Rates We used as a starting point the information on the number of angioplasty procedures performed in Ontario for 2004/2005, which is available from the Institute for Clinical Evaluative Services (ICES), on a sex-distinct and age-banded basis (data source 1). The rates per 100,000 of population 20 or older are:

Rates of Angioplasty per 100,000 in Population 20 or Older in Ontario, 2004/2005

Age Male Female 20–39 15 2 40–64 362 89 65–74 786 344 75+ 650 312

Age-standardized rate for male and female combined: 185.

We then interpolated/extrapolated these raw incidence rates over the full age range using a three-point Lagrange interpolation formula. We also performed a test using the 2004 Ontario population by age to see if the weighted average rate was equal to the interval raw incidence rates, and made some adjustments to the interval midpoints to replicate those rates.

Finally, we adjusted the rates to convert the Ontario rates to Canada rates, using the Canadian Institute for Health Information (CIHI) health indicators and the data on age-standardized percutaneous coronary interventions (PCI) rate per 100,000 of population over 20 (data source 2). This data is available for each province. Canadian age standardized rates were calculated using Statistics Canada data on population by province. We then applied the ratio of Canada to Ontario to the angioplasty sex-distinct and age-banded rates shown above. The result is:


62

Age Standardized PCI Rate per 100,000 age 20 or Older in Canada

2003/2004 2004/2005 Ontario 148.2 164.1 Canada 166.9 176.5

So for 2004/2005, the Canada to Ontario ratio is 107.6%.

The resulting raw angioplasty rates are:

Rates of Angioplasty per 100,000 age 20 or older in Canada, 2004/2005

Age Male Female 20 0.0 0.0 30 2.0 0.3 40 99.2 12.2 50 373.0 63.3 60 692.4 232.1 70 861.0 385.3 80 734.0 372.3

4.2.3.4 Adjustment for Trend to 2008 The data require adjustment for trend from December 2004 (the midpoint of the study period for the base data) to January 2008.

We performed a logarithmic regression on the CIHI PCI Canada age-standardized rates from 2003/2004 to 2006/2007 to project the 2008 rate, and the result is a 5.9% increase to the rates in section 4.2.3.3.

4.2.3.5 Adjustment to Count Only First-ever Incidences There are limited data available to adjust incidence rates to a first-ever basis. According to “A Contemporary Overview of Percutaneous Coronary Interventions” (data source 4), 32% of patients undergoing a PCI had already undergone a previous one. The average age of patients in the study was 64.

In line with what was done for heart attack, to reflect that this number should vary by age, and that it is expected that patients undergoing angioplasty at younger ages will be at their first-ever, we used 32% for ages 45-plus only, and 0% below 40 (graded from 0 to 32% between 40 and 45).

4.2.3.6 Adjustment for Sudden Deaths There is no sudden-death adjustment required, since angioplasty is a medical procedure and all procedures would be recorded.

4.2.3.7 Adjustment to Remove Overlap with Other Critical Illnesses We determined that for the “cardiac conditions” the order of precedence would be heart attack, stroke, CABG, and angioplasty respectively, and that overlap adjustments would be made accordingly. So for angioplasty, adjustments for heart attack, stroke and CABG are required.


63

We did not find literature that linked stroke to angioplasty so did not make any adjustment for stroke. For heart attack and CABG, according to the JACC article (data source 4), 29% of patients undergoing a PCI had a previous heart attack and 18.5% had a prior CABG. We were however reluctant to use those figures given that the trend since 2000 has been going up for PCI and down for heart attack and CABG and wanted to reflect the recent trends. Based on the CCORT Canadian Cardiovascular Atlas (data source 7, chapter 10), 20% of patients suffering from an AMI had a PCI in the following year. Taking this into consideration resulted in 19% instead of 29% of patients undergoing PCI would have had a previous heart attack. Using that ratio, and given that the trend on CABG is a third to a half of what it is for heart attack, we used 15% for CABG instead of the 18.5%. We adjusted for overlap between heart attack and CABG by using a 50% factor, consistent with the section on CABG. The final overlap adjustment is 26.5% (19% + 15% * 50%).

To determine the overlap rate by age, we ran a logarithmic regression on the sex distinct age banded revascularization rates published in “Revascularization and Heart Attack Outcomes” (data source 5) and applied that slope to the 26.5%. The result is as follows:

Age Male Female 40 and under 58.6% 46.7%

45 50.1% 40.1% 50 45.9% 36.8% 55 41.6% 33.5% 60 37.4% 30.2% 65 33.1% 26.8% 70 28.9% 23.5% 75 24.6% 20.2% 80 20.4% 16.9% 85 16.1% 13.6%

4.2.3.8 Adjustment for Prevalence Given the lack of published data on this, and the fact that the Staple Inn report used the same prevalence adjustment for angioplasty than for heart attack, we used the same adjustment that was used for heart attack. See the heart attack section for details.

4.2.3.9 Adjustment for the 30-day Survival Period We decided not to use the 30-day mortality data published for angioplasty, because we thought they would be greatly influenced by the heart attack overlap. Instead we used standard mortality, as detailed in section 3.8.


64


Note: The Derived Incidence Rate for Angioplasty =

(Raw Observed Rate) x (1 + Angioplasty Excl. from Definition Adjustment) x (1 + 2008 Trend Adjustment) x (1 + First-ever Incidence Adjustment) / (1 - Sudden Death Adjustment) x (1 + Overlap Adjustment) / (1 - Prevalence Adjustment) x (1 + 30-Day Survival Adjustment)


Adjustments:Angio Excl. from Definition 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%2008 Trend 5.9% 5.9% 5.9% 5.9% 5.9% 5.9%First Ever Incidence 0.0% -32.0% -32.0% 0.0% -32.0% -32.0%Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Overlap -58.6% -41.6% -24.6% -46.7% -33.5% -20.2%Prevalence 0.2% 3.5% 11.8% 0.0% 0.8% 4.9%30-Day Survival 0.0% 0.0% -0.3% 0.0% 0.0% -0.2%


Summary of Calculation of Incidence Rates per 1,000: AngioplastyMales Females


65


4.2.4 Heart Valve Replacement

4.2.4.1 Definition The CLHIA Benchmark Definition for heart valve replacement is as follows:

Heart valve replacement is defined as the undergoing of surgery to replace any heart valve with either a natural or mechanical valve. The surgery must be determined to be medically necessary by a specialist.

Exclusion: no critical illness benefit will be payable for heart valve repair.

A 30-day survival period is also representative.

This definition is consistent with the terms of contracts issued during the exposure period for the experience study, and was used to develop the incidence rates.

4.2.4.2 Data Sources 1. Hassan et al for the Canadian Cardiovascular Outcomes Research Team. “Use of

Valve Surgery in Canada”. Canadian Journal of Cardiology Vol. 20 No. 2 (2004): 149–154.

2. Thourani et al. “Ten-year Trends in Heart Valve Replacement Operations”. The Annals of Thoracic Surgery 70 (2000): 448–455.

3. Hassan et al for the Canadian Cardiovascular Outcomes Research Team. “Outcomes after Aortic and Mitral Valve Replacement Surgery in Canada”. Canadian Journal of Cardiology Vol. 20 No. 2 (2004): 155–163.

Angioplasty population incidence rates

0.0000

0.0010

0.0020

0.0030

0.0040

0.0050

0.0060

0.0070

0.0080

0.0090

18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99 Age

Rate Raw Male Raw Female Derived Male Derived Female


66

4. Manuel et al. “How Many People have had a Myocardial Infarction? Prevalence Estimated Using Historical Hospital Data”. BMC Public Health 7 (2007): 174.

5. ICES. InTool interactive web-based reporting tool. Website www.ices.on.ca.

4.2.4.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates “Use of Valve Surgery in Canada” (data source 1) was used to provide initial Canadian procedure rates for aortic valve replacement (AVR) and mitral valve replacement (MVR). Cases with concomitant CABG surgery were not included.

Age Group

AVR (rate per 100,000)

MVR (rate per 100,000)

Male Female Male Female 20–49 4.4 1.7 1.5 2.2 50–64 19.7 10.7 8.9 12.9 65–74 40.3 27.7 15.8 22.9 75–105 33.8 22.5 9.5 10.1

We then adjusted the rates to remove double counting for simultaneous AVR and MVR procedures, using statistics from “Ten-year Trends in Heart Valve Replacement Operations” (data source 2).

Age Group

AVR + MVR (procedures per

1,000) Male Female

20–49 0.056 0.035 50–64 0.270 0.214 65–74 0.533 0.466 75–105 0.416 0.308

These raw incidence rates have been interpolated/extrapolated over the full age range using a three-point Lagrange interpolation formula.

Adjustment for Heart Valve Replacement Excluded from the Definition No adjustment was required.

4.2.4.4 Trends The data require adjustment for trend from July 1997 (the midpoint of the study period for the base data) to January 2008. Information from “Use of Valve Surgery in Canada” (data source 1) provided trends from 1994/1995 to 1999/2000, and these were extrapolated using logarithmic regression, resulting in a 27% increase due to trend.

4.2.4.5 First-ever Adjustment “Use of Valve Surgery in Canada” (data source 1) provided clinical statistics showing 1.5% of valve surgeries were repeat procedures. This figure is used for all ages.

4.2.4.6 Sudden Death Adjustment There is no sudden-death adjustment required for procedure-based conditions.

http://www.ices.on.ca/�


67

4.2.4.7 Overlap Adjustment “Use of Valve Surgery in Canada” (data source 1) provided clinical statistics showing 3.1% of valve surgeries were performed after a recent heart attack. This figure is used for all ages. Although concomitant CABG is prominent, the data categorized valve surgery as standalone or accompanied by CABG, and we selected standalone cases.

4.2.4.8 Prevalence Adjustment Given the lack of published data on this and since the Staple Inn report used the same prevalence adjustment for angioplasty as for heart attack, we used the same adjustment that was used for heart attack. See the heart attack section for details.

4.2.4.9 Thirty-day Survival Adjustment Thirty-day survival was estimated from the paper “Outcomes after Aortic and Mitral Valve Replacement Surgery in Canada” (data source 3). The paper provided in-hospital mortality rates of 3.7% and 5.7% for AVR and MVR, respectively. Mortality rates by age were then obtained using estimated odds ratios from risk-adjustment models in the paper. Estimated case counts for AVR and MVR by age and gender were then used to calculate mortality for all valve surgeries by age and gender.

Age Group

Estimated In-Hospital Mortality

M F 20–64 2.9% 3.3% 65–74 5.0% 5.5% 75+ 6.9% 7.4%

Since no trend was evident for the in-hospital mortality data, no trend adjustment was applied. We used a Lagrange interpolation to adjust the rates for the incidence basis.


68


Summary of Calculation of Incidence Rates per 1,000: Heart Valve Replacement Males Females Central Age 35 55 75 35 55 75 Raw Observed Rate 0.04 0.24 0.52 0.02 0.18 0.44 Adjustments:

Surgeries Excl. from Definition 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

2008 Trend 27.4% 27.4% 27.4% 27.4% 27.4% 27.4%

First-ever Incidence -1.5% -1.5% -1.5% -1.5% -1.5% -1.5%

Sudden Death 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%

Overlap -3.1% -3.1% -3.1% -3.1% -3.1% -3.1% Prevalence 0.2% 3.5% 11.8% 0.0% 0.8% 4.9% 30-Day Survival -1.9% -3.4% -5.9% -2.3% -3.8% -6.3%



0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

Rat

e pe

r 1,0

00

Age Last

Heart Valve Replacement Incidence Rates Raw vs Derived



69

4.2.5 Aortic Surgery 4.2.5.1 Definition The CLHIA Benchmark Definition for aortic surgery is as follows:

Aortic surgery is defined as the undergoing of surgery for disease of the aorta requiring excision and surgical replacement of the diseased aorta with a graft. Aorta refers to the thoracic and abdominal aorta but not its branches. The surgery must be determined to be medically necessary by a specialist.

A 30-day survival period is also typical.

This definition is consistent with the terms of contracts issued during the exposure period for the experience study, and was used to develop the incidence rates.

4.2.5.2 Data Sources 1. Forbes et al. “National Audit of the Recent Utilization of Endovascular

Abdominal Aortic Aneurysm Repair in Canada: 2003 to 2004”. Journal of Vascular Surgery Vol. 42 No. 3.

2. Dueck et al. “Survival after Ruptured Abdominal Aortic Aneurysm: Effect of Patient, Surgeon, and Hospital Factors”. Journal of Vascular Surgery Vol. 42 No. 3.

3. Wen et al. “Hospital Volume, Calendar Age, and Short-term Outcomes in Patients Undergoing Repair of Abdominal Aortic Aneurysms: the Ontario Experience, 1988–92”. Journal of Epidemiology and Community Health 50 (1996): 207–213.

4. Bowen et al. Systematic Review and Cost-effectiveness Analysis of Elective Endovascular Repair Compared to Open Surgical Repair of Abdominal Aortic Aneurysms. Interim report for the Ontario Ministry of Health and Long-term Care. Program for Assessment of Technology in Health, McMaster University. Hamilton, ON. July 2005.

5. Katz et al. “Gender Differences in Abdominal Aneurysm Prevalence, Treatment, and Outcome”. Journal Of Vascular Surgery Vol. 25 No. 3.

6. Prinssen et al. “A Randomized Trial Comparing Conventional and Endovascular Repair of Abdominal Aortic Aneurysms”. The New England Journal of Medicine Vol. 351 No. 16.

7. The Information Centre. HES Online Hospital Episode Statistics. Self-service query.

8. Paterson, J.M., J.E. Hux, J.V. Tu and A. Laupacis. The Ontario Wait Time Strategy: No Evidence of an Adverse Impact on Other Surgeries. ICES Investigative Report. Institute for Clinical Evaluative Sciences. Toronto, 2007.

4.2.5.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates There was limited published Canadian information on aortic surgeries with the detail necessary to provide incidence rates by age and gender. However, since the incidence rates are relatively low for most ages, a variety of sources were used to provide reasonable estimates. Data from “Survival after Ruptured Abdominal Aortic Aneurysm: Effect of Patient, Surgeon, and Hospital Factors” (data source 2) provided the total


70

number of abdominal aortic aneurysm (AAA) repair procedures in Ontario from 1992 to 2001, for both ruptured and elective aneurysm procedures. The definition above includes procedures which are not covered by these statistics. A review of data from England concluded that the rates be increased by approximately 10% to include other procedures. “Hospital Volume, Calendar Age, and Short-term Outcomes in Patients undergoing Repair of Abdominal Aortic Aneurysms: the Ontario Experience, 1988–92” (data source 3) provided a breakdown of procedures by age bands. Data from England were used to estimate the procedure rate for males and females at different age bands. It should be noted that the incidence rates derived from the English data were much lower than the Canadian data suggested, and accordingly English data were only used to establish relative incidence levels. The gender relationship for Ontario procedures was broadly corroborated by the paper Systematic Review and Cost-effectiveness Analysis of Elective Endovascular Repair compared to Open Surgical Repair of Abdominal Aortic Aneurysms (data source 4). Finally, we interpolated/extrapolated the rates over the full age range using a three-point Lagrange interpolation formula.

Age

Raw Incidence Rates (Per

1,000) Male Female

30 0.001 0.000 40 0.003 0.001 50 0.028 0.006 60 0.791 0.151 70 2.247 0.421 80 2.275 0.498

Adjustment for Aortic Surgeries Excluded from the Definition Adjustments were made in the derivation of the raw rates to conform to the definition.

4.2.5.4 Trends The data require adjustment for trend from 1996 (the midpoint of the study period for the base data) to January 2008. The paper The Ontario Wait Time Strategy: No Evidence of an Adverse Impact on Other Surgeries (data source 8) provided data on elective AAA procedures performed in Ontario, which suggested the rate had fallen from approximately 16/100,000 in 1996 to 13/100,000 in 2005. The trend was extrapolated to 2008 using the average compound growth rate over the 1996 to 2005 period. This provides an adjustment of -23.3% which was applied at all ages and genders.

4.2.5.5 First-ever Adjustment The paper “A Randomized Trial Comparing Conventional and Endovascular Repair of Abdominal Aortic Aneurysms” (data source 6) described a study in which approximately 30% of elective AAA procedures were repeat procedures. It was assumed that, due to their nature, emergency AAA procedures would be first-ever. Multiplying these percentages by the elective and emergency procedures considered in the raw data implies


71

76% of the aggregate procedures are first-ever. This assumption was applied to all ages and genders.

4.2.5.6 Sudden Death Adjustment There is no sudden-death adjustment required for procedure-based conditions.

4.2.5.7 Overlap Adjustment The paper Systematic Review and Cost-effectiveness Analysis of Elective Endovascular Repair Compared to Open Surgical Repair of Abdominal Aortic Aneurysms (data source 4) estimated that myocardial infarction was an early complication following elective AAA repair in 2% to 8% of cases. Based on this, we have assumed that there is a 5% overlap for other conditions.

4.2.5.8 Prevalence Adjustment Given the lack of published data on this, and since the Staple Inn report used the same prevalence adjustment for angioplasty as for heart attack, we also used the same adjustment as for heart attack. See the heart attack section for details.

4.2.5.9 Thirty-day Survival Adjustment We estimated 30-day survival from the paper “Survival after Ruptured Abdominal Aortic Aneurysm: Effect of Patient, Surgeon, and Hospital Factors” (data source 2). The paper described 30-day mortality rates of 4.5% for elective AAA repair and 40.8% for ruptured AAA repair. Data from “Gender Differences in Abdominal Aneurysm Prevalence, Treatment, and Outcome” (data source 5) were used to adjust these rates by age and gender, using the Ontario procedure data from the Journal of Vascular Surgery (data source 2). No trend adjustment was applied.

Lagrange interpolation was then used to adjust this for our incidence basis.

30-Day

Mortality Age Male Female

20 4.90% 6.00% 25 4.90% 6.00% 30 4.90% 6.00% 35 4.90% 6.00% 40 4.90% 6.00% 45 4.90% 6.20% 50 4.90% 6.50% 55 4.90% 6.80% 60 4.90% 7.20% 65 6.80% 9.30% 70 9.00% 11.50% 75 12.10% 14.70% 80 17.50% 20.10% 85 25.10% 28.20% 90 49.80% 52.60%


72


Summary of Calculation of Incidence Rates per 1,000: Aortic Surgery Males Females Central Age 35 55 75 35 55 75 Raw Observed Rate 0.00 0.28 2.56 0.00 0.06 0.50 Adjustments Procedures Excl. from Definition 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

2008 Trend -23.3%

-23.3%

-23.3%

-23.3%

-23.3% -23.3%

First-ever Incidence -24.0%

-24.0%

-24.0%

-24.0%

-24.0% -24.0%

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap -5.0% -5.0% -5.0% -5.0% -5.0% -5.0% Prevalence 0.2% 3.5% 11.8% 0.0% 0.8% 4.9%

30-Day Survival -4.9% -4.9% -12.1% -6.0% -6.8% -14.7%



73


0,0

0,5

1,0

1,5

2,0

2,5

3,0

Rat

e pe

r 1,0

00

Age Last

Aorta Surgery Incidence Rates Raw vs Derived



74

4.3 Stroke 4.3.1 Definition The CLHIA Benchmark Definition which is representative of those used in fully-underwritten non-cancellable Canadian critical illness policies in 2008 is as follows:

Stroke (cerebrovascular accident) is defined as “a definite diagnosis of an acute cerebrovascular event caused by intra-cranial thrombosis or haemorrhage, or embolism from an extra-cranial source, with:

• Acute onset of new neurological symptoms, and

• New objective neurological deficits on clinical examination persisting for more than 30 days following the date of diagnosis. These new symptoms and deficits must be corroborated by diagnostic imaging testing. The diagnosis of stroke must be made by a specialist.

Exclusion: no benefit will be payable under this condition for:

• Transient ischemic attacks; or,

• Intra-cerebral vascular events due to trauma; or,

• Lacunar infarcts which do not meet the definition of stroke as described above.” The insured person must survive for 30 days following the date of diagnosis.

4.3.2 Data Sources 1. Canadian Institute for Health Information. Special request (Ontario Information for

period from 2004 to 2006). 2. Surveillance Division, Centre for Chronic Disease Prevention and Control. Chronic

Disease Infobase. 3. Statistics Canada. Population Projections for Canada, Provinces and Territories,

2005–2031. Catalogue no. 91-520-XIE. 4. Statistics Canada. 2005 Canadian Community Health Survey Public Use Metafile

(Cycle 3.1). 5. Tu, J., and J. Porter. Stroke Care in Ontario: Hospital Survey Results. 1999. 6. Holroyd-Leduc, J., M. Kapral, P. Austin, and J. Tu. Sex Differences and Similarities

in the Management and Outcome of Stroke Patients. 2000. 7. Lamassa et al. “Characteristics, Outcome, and Care of Stroke Associated With Atrial

Fibrillation in Europe: Data From a Multicenter Multinational Hospital-Based Registry (The European Community Stroke Project)”. Stroke 32 (2001): 392–398.

8. Estimate of Annual Stroke Incidence in England and Wales. www.stroke.org.uk/document.rm?id=732

9. American Heart Association. Heart Disease and Stroke Statistics, 2008. 10. Canadian Institute for Health Information. Health Care in Canada 2007. 11. Dinani et al. A Critical Review. Staple Inn Actuarial Society, 2000. 12. Heart and Stroke Foundation of Canada. The Growing Burden of Heart Disease and

Stroke in Canada 2003. 13. Robjohns et al. Exploring the Critical Path. Staple Inn Actuarial Society, 2006.

http://www.stroke.org.uk/document.rm?id=732�


75

4.3.3 Calculation of Base Incidence Rates The general approach set out in section 3 was followed. Points specific to the development of incidence rates for stroke are addressed below.

Raw Observed Incidence Rates In Canada the Canadian Institute for Health Information (CIHI) compiles statistics for each province on hospital discharges, but a lag exists between the date the information is gathered and the date the information is released. We requested a special tabulation of all discharges for ICD codes I-60 to I-66 inclusive from the CIHI for data from 2004 through 2006. It was decided to include these categories by consultation with medical experts as it seemed to best fit the definition. Transient ischemic attacks (TIAs) are specifically excluded from these categories and as such it was decided that with inclusion of the categories below no other adjustment would be needed to reflect differences between the data and the underlying definition:

I-60: Subarachnoid hemorrhage;

I-61: Intra-cerebral hemorrhage;

I-62: Other non-traumatic intracranial hemorrhage;

I-63: Cerebral infarction;

I-64: Stroke, not specified as hemorrhage or infarction;

I-65: Occlusion and stenosis of pre-cerebral arteries, not resulting in cerebral infarction; and

I-66: Occlusion and stenosis of cerebral arteries, not resulting in cerebral infarction.

To reduce the cost we confined the request to Ontario, the most populous province in Canada, and then made subsequent adjustments to provide Canadian incidence rates.

CIHI provided data for all stroke cases, those where stroke was the most responsible diagnosis, and split these into acute care admissions, emergency room admissions, and day surgery admissions. We chose the data for acute care admissions, where stroke was the most responsible diagnosis, as best representing the definition shown above, and this basis is fairly consistent with the stroke statistics reported in Canada in such publications as The Growing Burden of Heart Disease and Stroke in Canada 2003.

We decided to use the most recent data year available, which was 2006. We then divided the number of discharges for 2006 by the estimated population of Ontario in 2006 to provide crude rates. We used the Statistics Canada population estimate rather than the census figure, since the census figure does not include a significant portion of the population which is not enumerated.


76

Distribution of Stroke by Age Group and Gender, Ontario, 2006

Age Group

2006 New Case Estimates*

2006 Population Estimates Raw incidence Rates

(in 1,000s) (per 1,000) Male Female Total Male Female Total Male Female Total

0–19 53 32 85 1,595 1,523 3,118 0.0332 0.0210 0.0273 20–24 17 23 40 444 430 874 0.0383 0.0535 0.0458 25–29 23 34 57 424 425 849 0.0542 0.0800 0.0671 30–34 61 39 100 446 446 892 0.1368 0.0874 0.1121 35–39 62 76 138 480 483 963 0.1292 0.1573 0.1433 40–44 171 144 315 546 541 1,087 0.3132 0.2662 0.2898 45–49 331 246 577 511 514 1,025 0.6477 0.4786 0.5629 50–54 424 288 712 434 448 882 0.9770 0.6429 0.8073 55–59 659 392 1,051 383 397 780 1.7206 0.9874 1.3474 60–64 841 524 1,365 287 299 586 2.9303 1.7525 2.3294 65–69 1,059 680 1,739 223 243 466 4.7489 2.7984 3.7318 70–74 1,293 883 2,176 188 213 401 6.8777 4.1455 5.4264 75–79 1,429 1,266 2,695 151 189 340 9.4636 6.6984 7.9265 80–84 1,331 1,636 2,967 98 151 249 13.5816 10.8344 11.9157 85-89 787 1,337 2,124 44 84 128 17.8864 15.9167 16.5938 90+ 342 798 1,140 17 47 64 20.1176 16.9787 17.8125

Total 8,883 8,398 17,281 6,271 6,433 12,704 1.4165 1.3055 1.3603

* Counts of hospitalizations with a Most Responsible Diagnosis, includes all of ICD codes I60 to I66.

Ontario to Canada Adjustment We compared information from the Chronic Disease Infobase with the age-standardized incidence rates for Ontario to those of Canada. The ratio of Canada to Ontario for males is 100.4% using a weighted average approach giving higher weight to more recent years. This ratio was observed over the calendar year period 1994 to 2005. For females the ratio is 98.3% using the same method. Since the ratios were in close to 100%, we decided to not make an adjustment. That is, Ontario appropriately reflects the rates for Canada as a whole.

4.3.4 Trends The data require adjustment for trend from October 2006 (the midpoint of the study period for the base data) to January 2008. The incidence curve has decreased sharply over the past five years, which would normally suggest the trend adjustment should reduce the incidence rate. In some jurisdictions events such as smoking bans have led to one-time immediate decreases in stroke incidents. Ontario has led an aggressive campaign against smoking over the past few years, which resulted in a significant reduction in smoking prevalence in 2005 and 2006, but early indications show Ontario smoking prevalence increased in 2007. There has also been progress in education regarding cardiovascular health, and increased monitoring for risk factors—this may have the effect of deferring strokes rather than eliminating them altogether. We are also faced with the increase in diabetes and obesity, which will tend to increase stroke incidence rates in the long term. Based on all of these


77

factors, we decided not to adjust the incidence rates for the mathematically indicated trend adjustment.

The sharp reduction in incidence discussed above led to debate among the members of the subcommittee regarding the accuracy of the data and their application in the context of developing a population incidence table for critical illness. The data come from the best Canadian source, and we have verified that the base information is consistent with prior reported periods. It will be important to monitor this closely over the next few years to understand the forces behind these reductions, and their relationship to critical illness incidence rates.

We used a Lagrange interpolation to smooth the incidence rates, and to provide other interpolations used below.

4.3.5 First-ever Adjustment The report Heart Disease and Stroke Statistics published by the American Heart Association in 2008 estimated that out of 780,000 strokes in the U.S., 600,000 would be first attacks and 180,000 would be recurrent attacks. A UK source also found which indicated an overall recurrence level of 40%. We thought that the U.S. overall recurrence rate would likely be a better approximation for the level of recurrence in Canada. The UK source was useful in providing finer breakdown by gender and age bracket which we used to split the overall rate of 23% (180,000/780,000) among the various subsets. The first-ever adjustment for females is 80% up to age 70 and 77% at ages 80+ with linear interpolation between. For males it is 83% up to age 55, 72% at age 70, and 74% at ages 80+ with linear interpolation between these ages.

First-ever Adjustment Factors

Age Group Male Female <=55 83% 80%

…

…

… 70 72% 80%

…

…

…

>=80 74% 77%

4.3.6 Sudden Death Adjustment There is no sudden-death adjustment required, since we have used only acute-care discharge data for those who were admitted to hospital.

4.3.7 Overlap Adjustment We assumed that for the “cardiac conditions” the order of precedence is heart attack, stroke, CABG, and angioplasty respectively, and that overlap adjustments would be made accordingly. No adjustment is required for heart attack. For stroke, an adjustment for heart attacks will be required, while adjustments for heart attack and stroke will be required for CABG. The adjustment for stroke overlap with heart attack is to remove 7% of the incidence. This figure was taken from the paper Exploring The Critical Path produced for the Staple Inn Actuarial Society.


78

4.3.8 Prevalence Adjustment We estimated prevalence using figures found in the Statistics Canada 2005 Canadian Community Health Survey. These prevalence figures were provided by gender for various age groupings. We used a Lagrange interpolation to adjust these to fit all ages.

Prevalence Adjustment Factors

Age Group Male Female

20–29 0.2% 0.1% 30–39 0.3% 0.3% 40–49 0.4% 0.6% 50–59 1.2% 0.9% 60–69 3.0% 1.9% 70–79 4.6% 3.4% 80+ 8.5% 6.8%

4.3.9 Thirty-day Survival Adjustment We estimated the 30-day survival using data presented in Stroke Care in Ontario: Hospital Survey Results. The data on 30-day mortality was provided for 1992 through 1996 by the age groupings 20–64, 65–74, and 75-plus and split by gender. We compared these rates with those of 2004 for appropriateness. We set the rates to appropriate age pivot points in the groupings using incidence by age as a weight and applied a Lagrange interpolation between these pivot points.

Thirty-day Mortality Factors (Deaths per 100,000)

Age Group Male Female 20–64 10.8 11.0 65–74 15.4 15.0 75–84 22.4 20.4 85+ 33.0 31.3


79

4.3.10 Summary of Rates and Adjustments

Summary of Calculation of Incidence Rates per 1,000: Stroke Males Females Central Age 35 55 75 35 55 75 Raw Observed Rate 0.1439 1.4088 8.3198 0.1316 0.8523 5.6201 Adjustments: Stroke Excl. from Definition 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

Ontario to Canada 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence -17.0% -17.0% -26.9% -20.0% -20.0% -21.5% Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap -7.0% -7.0% -7.0% -7.0% -7.0% -7.0% Prevalence 0.3% 1.2% 4.6% 0.3% 0.9% 3.4% 30-Day Survival -9.8% -10.8% -18.8% -9.1% -11.1% -17.5% Derived Incidence Rate 0.1004 0.9816 4.8156 0.0892 0.5691 3.5050


0,0

2,0

4,0

6,0

8,0

10,0

12,0

14,0

16,0

18,0

15

19

23

27

31

35

39

43

47

51

55

59

63

67

71

75

79

83

Rat

e pe

r 1,0

00

Age Last

Stroke Incidence Rates Raw vs Derived



80

4.4 Kidney Failure 4.4.1 Definition The CLHIA Benchmark Definition is as follows:

Kidney failure is defined as “a definite diagnosis of chronic irreversible failure of both kidneys to function, as a result of which regular haemodialysis, peritoneal dialysis or renal transplantation is initiated. The diagnosis of kidney Failure must be made by a specialist”.


A specialist is a licensed medical practitioner who has been trained in the specific area of medicine relevant to the covered critical illness condition for which benefit is being claimed, and who has been certified by a specialty examining board. In the absence or unavailability of a specialist, and as approved by the insurer, a condition may be diagnosed by a qualified medical practitioner practising in Canada or the U.S.


4.4.2 Data Sources 1. Canadian Institute for Health Information (CIHI). Quick Stats: New End-stage Renal

Disease (ESRD) Patients by Age Group and Sex at Start of Treatment, Canada, 2001–2005. Website: www.cihi.ca.

2. Statistics Canada. Annual Demographic Estimates: Canada, Provinces and Territories 2007—Revised. Catalogue no. 91-215-4.

3. The Renal Association: UK Renal Registry. The Fourth Annual Report. December 2001 (Chapter 1, Page 1).

4. ———. The Tenth Annual Report. December 2007 (Chapter 1, Page 1). 5. Ianhez et al. “Renal Cell Carcinoma in Renal Transplant Patients”. UroToday, July 2,

2007. 6. The Staple Inn Actuarial Society. A Critical Review: Report of the Critical Illness

Healthcare Study Group. March 14, 2000 (Page 37). 7. CIHI. Quick Stats: Final Treatment Modality for End-stage Renal Disease (ESRD)

Patients in CORR. December 31, 2005.

4.4.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates The Canadian Institute for Health Information (CIHI) has statistics available on its website for calendar years 1996 through 2005 (data source 1). The data were collected for end-stage renal disease (ESRD), which is nearly synonymous with the critical illness definition of “chronic irreversible failure of both kidneys to function”. The exception is situations where there is non-chronic ESRD, such as caused by poison. We made no adjustment.

CIHI only collects data in situations where treatment is acquired, but because the critical illness definition requires treatment there is alignment between data and definition. The same three treatments apply in both the data and definition, namely haemodialysis, peritoneal dialysis and renal transplantation.

http://www.cihi.ca/�


81

The data from CIHI are for “new” ESRD patients and are therefore considered first-incidence.

Although we could have acquired CIHI data from 2006 and 2007 for a cost, there has been no material change in the definition of ESRD nor is the trend in new ESRD patients overly volatile. We therefore decided that 2005 data trended forward to 2008 were suitable.

We then divided the incidence by the estimated population to produce crude rates. We used the Statistics Canada population estimate rather than the census figure, since the census figure does not include a significant portion of the population which is not enumerated.

New End-stage Renal Disease (ESRD) Patients by Age Group and Sex at Start of Treatment, Canada, 2005 (data source 1)

Population Estimates (000’s)

2005 data New ESRD Patients

2005 data Raw incidence (per

1000)

Age Assumed

mid-pt Male Female Total Male Female Total Male Female Total 0–17 8.5 3,576 3,406 6,982 47 29 76 0.013 0.009 0.011 18–44 31 6,401 6,251 12,653 381 236 617 0.060 0.038 0.049 45–64 54.5 4,188 4,264 8,452 1,038 627 1,665 0.248 0.147 0.197 65–74 69.5 1,065 1,174 2,239 813 553 1,366 0.763 0.471 0.610 75+ 80 774 1,213 1,987 854 622 1,476 1.103 0.513 0.743

We used a three-point Lagrange interpolation, with the midpoint of the age groups above, to determine an incidence rate for each age. This created a decreasing incidence rate for older females which was not consistent with the raw incidence, so we made two adjustments to the interpolated rates:

1. We interpolated female rates linearly between ages 70 and 80; and 2. We set female rates above age 80 to equal the same percentage of male rates as at age

80.

Kidney failure incidence rates (after adjustments)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 94 98

Age

Per 1

000

malefemale


82

Adjustment for Kidney Failure Excluded from the Definition As mentioned above, the data was collected for end-stage renal disease, which is nearly synonymous with the critical illness definition of “chronic irreversible failure of both kidneys to function”. The exception is situations where there is non-chronic ESRD. We believe the incidence of such is immaterial.

4.4.4 Trends The data require adjustment for trend from July 2005 (the midpoint of the data) to January 2008. We used the same data as in section 3.1 to create the raw incidence rates for 2001 through 2005.

Raw Incidence Rates for New ESRD Patients, Age-standardized to 2005 Population

Year Male Female 2001 0.200 0.139 2002 0.192 0.140 2003 0.201 0.130 2004 0.195 0.133 2005 0.196 0.127

Average annual trend

-0.5% -2.3%

Trend is required for 2.5 years, so we compounded the annual factors for 2.5 years to get a trend factor of -1.4% for males and -5.7% for females.

4.4.5 First-ever Adjustment The CIHI data are for “new” ESRD patients and are therefore aligned with our requirement for first-incidence. Therefore no adjustment is required for first-ever incidence.

4.4.6 Sudden Death Adjustment No sudden-death adjustment is required, since the CIHI data includes only patients that have acquired treatment for ESRD and have therefore survived long enough to receive treatment.

4.4.7 Overlap Adjustment The Staple Inn report (data source 6) indicates that renal failure overall has a significant correlation with other critical illnesses. However, the definition rests on acceptance for dialysis or transplantation, and such treatment is unlikely to be given where renal failure is secondary to some other disease. The report indicates that using incidence of treatment already makes a reasonable allowance for any overlap with other critical illnesses.

However, there is some evidence to suggest that the other critical illnesses, which may be the cause of kidney failure, are not diagnosed until after kidney treatment has commenced. The above-mentioned study published in UroToday (data source 5) in 2007 examined the records of 1,375 consecutive renal transplant patients and found that 1.37% of patients had renal cell carcinoma.

Some adjustment for overlap seems appropriate. Due to the lack of other collaborating studies and the limited scope of the UroToday study, we used a conservative 1% overlap assumption.


83

4.4.8 Prevalence Adjustment To account for people living with transplanted kidneys or on dialysis we used a CIHI data source.

Final Treatment Modality for End-Stage Renal Disease (ESRD) Patients in CORR on December 31, 2005(7)

Dialysis Transplant TOTAL 0–54 6,019 7,336 13,354 55–64 3,345 3,168 6,512 65–74 4,857 1,738 6,595 75+ 5,520 432 5,952

Furthermore, the prevalence varies by gender as demonstrated by the statistics quoted from the Renal Association (data source 4):

Age group Mid-age Male/Female Prevalence Ratio

0–54 27 1.500 55–64 59.5 1.500 65–74 69.5 1.800 75–79 77 2.300 80–84 82 2.900 85–90 87.5 4.600 91+ 93 7.900

To combine these three sources of data, these steps were followed:

1. We developed a prevalence rate by age group by dividing the TOTAL column in the first table by the 2005 population estimate. We interpolated these prevalence rates using a three-point Lagrange formula. The rates for ages 0 to 30 were linearly interpolated from zero to the age 30 rate.

2. We interpolated the Male/Female Prevalence Ratio from the second table using a three-point Lagrange formula to determine a ratio for all ages, keeping the ratio at 1.5 for all ages under 60 and 7.9 for all ages 93-plus.

3. We used the two sets of interpolated rates from 1. and 2. to develop gender-specific rates for all ages.


84

There is no adjustment to trend the prevalence to 2008.

4.4.9 Thirty-day Survival Adjustment The Renal Association published one-year survival rates by age-band in its 2001 report (data source 3) . Its 2007 report (data source 4) confirmed the overall survival rate has not changed materially, though it does not provide a split by age. We converted the 2001 rates to annual mortality rates and compared them to an estimate of general population mortality (72.3% and 80.3% of the CIA86-92 aggregate table for males and females respectively). We applied a 75% factor to the resulting multiples for each age-band to add conservatism to adjust for any differences in UK survival rates and Canadian survival rates, as well as any differences in mortality during the first 30-days compared to the annual survival rate.

Age band

One-year survival

rate (2001 study)

Multiple of general

population mortality

Multiple used to

determine 30-day survival

18–34 96% 80.84 60 35–44 94% 62.18 46 45–54 90% 42.10 31 55–64 84% 25.36 19 65–74 72% 16.47 12 85+ 65% 2.98 2

We applied the numbers in the right column above to the general population mortality (72.3% and 80.3% of the CIA86–92 aggregate table for males and females respectively) to determine the 30-day survival factor.

Kidney Failure Prevalence

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 94 98

Age

per 1

000

MaleFemale


85


2008 Canadian Kidney Failure Incidence Rate (per 1000) Males Females Central Age 35 55 75 35 55 75 Interpolated Raw Observed Rate 0.081 0.260 0.944 0.051 0.155 0.495 Adjustments: 2008 Trend -1.35% -1.35% -1.35% -5.75% -5.75% -5.75% Overlap -1.00% -1.00% -1.00% -1.00% -1.00% -1.00% Prevalence -0.07% -0.17% -0.42% -0.04% -0.12% -0.20% 30-Day Survival -0.51% -1.44% -2.70% -0.51% -1.44% -2.70% Derived Incidence Rate 0.079 0.250 0.893 0.047 0.142 0.448

4.5 Major Organ Transplant and Major Organ Failure on Waiting List 4.5.1 Definition The definitions, which are representative of those used in fully-underwritten non-cancellable Canadian critical illness policies in January 2008, are as follows:

• Major organ transplant is defined as “a definite diagnosis of the irreversible failure of the heart, both lungs, liver, both kidneys or bone marrow, and transplantation must be medically necessary. To qualify under major organ transplant, the insured person must undergo a transplantation procedure as the recipient of a heart, lung, liver, kidney or bone marrow, and limited to these entities. The diagnosis of the major organ failure must be made by a specialist.”

• Major organ failure on waiting list is defined as “a definite diagnosis of the irreversible failure of the heart, both lungs, liver, both kidneys or bone marrow, and

Kidney Failure Incidence

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

15 19 23 27 31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95 99

Age

per 1

000

MaleFemale


86

transplantation must be medically necessary. To qualify under major organ failure on waiting list, the insured person must become enrolled as the recipient in a recognized transplant centre in Canada or the United States of America that performs the required form of transplant surgery. For the purposes of the survival period, the date of diagnosis is the date of the insured person’s enrolment in the transplant centre. The diagnosis of the major organ failure must be made by a specialist.”


A specialist is a licensed medical practitioner who has been trained in the specific area of medicine relevant to the covered critical illness condition for which benefit is being claimed, and who has been certified by a specialty examining board. In the absence or unavailability of a specialist, and as approved by the insurer, a condition may be diagnosed by a qualified medical practitioner practising in Canada or the U.S.


4.5.2 Data Sources 1. Canadian Institute for Health Information (CIHI). Quick Stats: Transplants by Organ

Type, Canada, 2001–2005. Website: www.cihi.ca. 2. Statistics Canada. Annual Demographic Estimates: Canada, Provinces and

Territories, 2007—Revised. Catalogue no. 91-215-X. 3. CIHI. E-statistics Report on Transplant, Waiting List and Donor Statistics: Table 2B,

2005 Summary Statistics. 4. Johns Hopkins Medicine. “Excellent Survival Rates for Liver Cancer Patients

Undergoing Transplant”. Press release, October 30, 2003. See www.hopkinsmedicine.org/Press_releases/2003/10_30_03.html.

5. BC Transplant. Stats & FAQ’s: Time Waiting for Organ Transplant. Data retrieved from the BC transplant website on September 20, 2008. See www.transplant.bc.ca/stats_faqs_main.htm.

6. CIHI. E-statistics Report on Transplant, Waiting List and Donor Statistics, 2001 thru 2004 Summary Statistics.

4.5.3 Calculation of Base Incidence Rates Raw Observed Incidence Rates and Overlap We estimated the incidence of both major organ transplant and major organ failure on waiting list together. Organ transplants in Canada generally require that the recipient first join the waiting list for that organ. To estimate the incidence of these two covered conditions requires a three-step process:

1. Estimate the incidence of actual organ transplant; 2. Apply a factor to account for people who join a waiting list but subsequently never

receive an organ transplant; and 3. Accelerate the incidence by the average wait time on a waiting list to capture that

people will claim once they are on the waiting list and not wait until the actual transplant.

http://www.cihi.ca/�

http://www.hopkinsmedicine.org/Press_releases/2003/10_30_03.html�

http://www.transplant.bc.ca/stats_faqs_main.htm�


87

The Canadian Institute for Health Information (CIHI) had organ transplant statistics freely available on its website for calendar years 1996 through 2005 (data source 1). The data were collected for transplants of all the organs specified in the critical illness definition except bone marrow. We decided to include only the data for heart, liver, heart and lung, single lung, and double lung transplants. Kidney transplants are included in the Kidney Failure incidence and therefore need not be included. The vast majority of bone marrow transplants are a result of cancer, so only a minor adjustment is required to include the remaining bone marrow transplants in this incidence. We have assumed no overlap with heart attacks.

Furthermore, we believe transplants only occur in Canada when they are medically necessary, so no adjustment need be made for that aspect of the definition.

Although CIHI has data from 2006 and 2007, the relatively materiality of organ transplant on the overall critical illness incidence is very low. We therefore determined that 2005 data trended forward to 2008 were suitable and we believe would not cause undue errors in overall incidence estimates.

Liver, Heart, Heart-lung, Single Lung, Bilateral Lung Transplants, Canada, 2005(1)

Population Estimates (000’s)

2005 data Organ Transplants

2005 data* Raw incidence (per

1000) Age Male Female Total Male Female Total Male Female Total 0–17 3,576 3,406 6,982 43 46 89 0.012 0.014 0.013 18–44 6,401 6,251 12,653 97 78 175 0.015 0.012 0.014 45–64 4,188 4,264 8,452 285 130 515 0.068 0.030 0.049 65+ 1,065 1,174 2,239 31 12 43 0.017 0.005 0.010

* 3% of liver transplants have been removed from the incidence data to reflect the incidence of cancer within liver transplant patients. (Data source 4)

In 2005 the proportion of people who were removed from the waiting list, by death or otherwise, without an organ transplant was 34% (data sources 1 and 3).

Raw incidence (per 1000)

Proportion removed from

waiting list Adjusted incidence (per

1000)

Age Mid-

pt Male Female Total Total Male Female Total 0–17 8.5 0.012 0.014 0.013 34% 0.018 0.021 0.019 18–44 31 0.015 0.012 0.014 34% 0.023 0.019 0.021 45–64 54.5 0.068 0.030 0.049 34% 0.103 0.046 0.075 65+ 70 0.017 0.005 0.010 34% 0.026 0.008 0.015

We added 0.01 per 1000 to these adjusted incidence rates to account for non-cancer-related bone marrow transplant. An extensive online search revealed a handful of non-cancer-related illnesses that could be treated with bone marrow transplant (the most prevalent of which is aplastic anaemia), all of which had incidences below 0.01 per 1000 and would have an even lower incidence of treatment by bone marrow transplant. We consider 0.01 per 1000 to be a conservative estimate of the sum of these non-cancer related bone marrow transplants.


88

We used a Lagrange interpolation, using the midpoint of the age groups above to determine an incidence rate for each age between 31 and 80. Incidence rates under 31 were linearly interpolated and incidence was set to zero above age 80.

Finally, we adjusted all incidence rates downward one age to reflect the average time spent on a waiting list. BC Transplant had statistics on waiting list times going back to 1998 (data source 5) . These statistics suggest that waiting times will vary by organ and by year, but are generally two to 15 months. For conservatism and simplicity of implementation, we used a 12-month adjustment.

Adjustment for Incidences Excluded from the Definition We made no adjustment.

4.5.4 Trends The data require adjustment for trend from July 2005 (the midpoint of the data) to January 2008. Using incidence of transplant to develop trend could be misleading because changes in average wait list times would influence this trend. A more accurate indicator of trend comes from the new waiting list entrants. CIHI publishes data on waiting list terminations (data source 6) as well as transplants (data source 1) which we combined to develop a waiting list movement report for heart, liver, heart and lung, single lung, and double lung transplants. We calculated the New Entrants column in the following table using the other five columns:

Waiting List Movement Summary BoY Withdrawals Deaths Transplants New Entrants EoY 2002 742 131 143 680 1,024 812 2003 812 153 159 669 1,041 872 2004 872 155 165 690 1,111 973 2005 973 149 211 733 1,168 1,048

Average Annual

Trend 4.5% 2.5 Year Trend 11.6%

We assumed that trend was equal for males and females. We did not adjust the waiting lists for deaths or withdrawals not yet reported to the organ transplant authorities.

4.5.5 First-ever Adjustment Data on the prevalence of people in the population with one or more transplanted organs are not readily available. We assumed that the prevalence is immaterial and made no adjustment for either first-ever incidence or for prevalence.

4.5.6 Sudden Death Adjustment There is no sudden-death adjustment required since the CIHI data include only patients that have survived long enough to be diagnosed and inserted into the organ transplant waiting list.

4.5.7 Overlap Adjustment Overlap considerations and adjustments are discussed in section 4.5.3.


89

4.5.8 Prevalence Adjustment Data on the prevalence of people in the population with one or more transplanted organs are not readily available. We assumed that the prevalence is immaterial and made no adjustment for either first-ever incidence or for prevalence.

4.5.9 Thirty-day Survival Adjustment The adjustment required for 30-day survival relates primarily to the mortality of those on the waiting list because it is unlikely that a patient would receive a transplant within 30 days of being put on the waiting list.

We used the data from the waiting list terminations (data source 3) to determine the average waiting list mortality rate in 2005 (21%, or 210 per 1000). We assumed that this mortality rate was uniform across ages and genders, converted it to a monthly mortality rate and applied it to the incidence.

Major Organ Transplant&

Major Organ Failure on Waiting ListIncidence

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90

Attained Age

per 1

000

MaleFemale


90


2008 Canadian Major Organ Transplant and Major Organ Transplant on Waiting List Incidence Rate (per 1,000)

Males Females Central Age 35 55 75 35 55 75 Interpolated Raw Observed Rate (after one-year shift due to waiting list)

0.044 0.110 0.013 0.032 0.055 0.006

Adjustments: 2008 Trend 11.59% 11.59% 11.59% 11.59% 11.59% 11.59% 30-Day Survival -1.92% -1.92% -1.92% -1.92% -1.92% -1.92% Derived Incidence Rate 0.048 0.121 0.014 0.035 0.060 0.007

4.6 Multiple Sclerosis 4.6.1 Definition The definition which is representative of those used in fully-underwritten non-cancellable Canadian critical illness policies in January 2008, is as follows:

Multiple sclerosis is defined as:

“a definite diagnosis of at least one of the following:

• Two or more separate clinical attacks, confirmed by magnetic resonance imaging (MRI) of the nervous system, showing multiple lesions of demyelination; or,

• Well-defined neurological abnormalities lasting more than six months, confirmed by MRI imaging of the nervous system, showing multiple lesions of demyelination; or,

• A single attack, confirmed by repeated MRI imaging of the nervous system, which shows multiple lesions of demyelination which have developed at intervals at least one month apart.

The diagnosis of multiple sclerosis must be made by a specialist.”


4.6.2 Data Sources 1. Hader, W.J., and I.M. Yee. “Incidence and Prevalence of Multiple Sclerosis in

Saskatoon, Saskatchewan”. Neurology 69 (2007): 1224–1229. American Academy of Neurology.

2. Larry Svenson. The Epidemiology of Multiple Sclerosis in Alberta. Presentation. Alberta Health and Wellness, 2004.

3. Robjohns et al. Exploring the Critical Path. Staple Inn Actuarial Society, 2006. 4. Dinani et al. A Critical Review. Staple Inn Actuarial Society, 2000.


91

4.6.3 Calculation of Base Incidence Rates The general approach set out in section 3 was followed. Points specific to the development of incidence rates for multiple sclerosis are addressed below.

Raw Observed Incidence Rates It was difficult to find sources with extensive data for the incidence of multiple sclerosis in Canada. One of the most current and complete studies which was performed in Canada and found by the committee was “Incidence and Prevalence of Multiple Sclerosis in Saskatoon, Saskatchewan” (data source 1). Given the availability of data within this article and the lack of other data it was decided to use this study as the basis for deriving incidence rates.

The following shows the incidence of multiple sclerosis in Saskatoon representative of 1970 to 2004:

Incidence of Multiple Sclerosis in Saskatoon, 1970 to 2004 (data source 1) Men Women Total Age Population Cases Rate* Population Cases Rate* Population Cases Rate* 0–14 20,115 - - 19,285 2 0.3 39,400 2 0.2 15–24 15,960 23 4.1 16,975 109 18.3 32,935 132 11.5 25–34 14,182 57 11.5 14,123 128 25.9 28,305 185 18.7 35–44 14,748 44 8.5 16,122 113 20.0 30,875 157 14.5 45–54 12,665 26 5.9 13,370 36 7.7 26,030 62 6.8 55–64 7,160 4 1.6 7,795 13 4.8 14,955 17 3.2 65–74 5,415 2 1.1 6,680 1 0.4 12,095 3 0.7 >=75 4,370 - - 7,845 - - 12,210 - - Total 94,615 156 4.7 102,195 402 11.2 196,815 558 8.1 * Incidence rate per 100,000 Saskatoon population, 2001.

Saskatoon to Canada Adjustment The results of this study were specific to Saskatoon, Saskatchewan, which has a different ethnic population mix than Canada as a whole. Specifically, Saskatoon has a higher proportion of aboriginal population which is suggested to have a lower incidence of multiple sclerosis (data source 2). Thus we decided that it would be appropriate to make an adjustment to the results above to reflect the different population mix of Saskatoon relative to Canada. The overall raw incidence rates were multiplied by 104.1% for males and 103.6% for females to better represent a Canadian population. We derived these rates by taking the proportion of aboriginals (including Metis) residing in Saskatoon relative to Canada and applying relative risk factors.

The estimated multiple sclerosis prevalence for Canada is 240 (data source 2) per 100,000. While this is lower than the prevalence of 298 (data source 1) in Saskatoon, Saskatchewan, we decided that considering the lack of data, no adjustment to incidence would be made.

Progressive Form Adjustment The insurance context definition for multiple sclerosis above pays out on a more progressive form of the condition than simply initial diagnosis of multiple sclerosis. The Saskatoon analysis is based on “clinical definite and probable cases” which we have assumed is


92

representative of the progressive form of multiple sclerosis. Since the data use initial diagnosis as the incidence of multiple sclerosis, a method is needed to extend out the dates of diagnosis in the context used in the study to that which would better represent a critical illness insurance context. The method chosen was similar to that which was used in the Staple Inn Actuarial Society report Exploring the Critical Path (data source 3). It assumes that 80% of all multiple sclerosis diagnoses convert to the progressive form at some point with 10% occurring immediately on initial diagnosis and the other 70% spread evenly over the following 14 years. The method we decided to use was similar to this but to not remove the 20% as it is believed that the data underlying the Saskatoon study would have already adjusted for this. Thus it is assumed that 10% of the underlying data would result in immediate payout and the other 90% are spread out evenly across the following 14 years.

We have not adjusted the incidence rates to reflect mortality among multiple sclerosis cases over the progressive duration. While this is similar to the Staple Inn Actuarial Society report cited above, the reader may wish to reflect this consideration in his/her own analysis.

Adjustment for Incidences Excluded from the Definition We made no adjustment.

4.6.4 Trends There is no conclusive evidence found suggesting an increasing or decreasing trend in the incidence of multiple sclerosis in Canada.

4.6.5 First-ever Adjustment There is no adjustment necessary to reflect a first-ever occurrence because multiple sclerosis is a progressive condition with symptoms that change over decades of the diagnosed person’s lifespan.

4.6.6 Sudden Death Adjustment There is no sudden-death adjustment required.

4.6.7 Overlap Adjustment There is no overlap adjustment required.

4.6.8 Prevalence Adjustment The prevalence of multiple sclerosis was determined using the same study as the one used for the raw incidence rates. Estimates suggest that 55,000 to 75,000 cases exist in Canada, which results in an adjustment which is very close to zero.

4.6.9 Thirty-day Survival Adjustment No 30-day survival adjustment required.


93


Summary of Calculation of Incidence Rates per 1,000: Multiple Sclerosis Males Females Central Age 35 45 55 35 45 55 Base Rate 0.0995 0.0741 0.0328 0.2375 0.1268 0.0644 Adjustments: Canada Adjustment 4.11% 4.11% 4.11% 3.56% 3.56% 3.56% Definition Adjustment 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Progressive Form Adjustment -3.77% 23.16% 86.81% 1.00% 60.40% 62.81%

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 30-Day Survival 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Derived Incidence Rate 0.0997 0.0950 0.0637 0.2484 0.2105 0.1085


94

Where the progressive form adjustment is a plus/minus:

Summary of Calculation of Incidence Rates per 1,000: Multiple Sclerosis Males Females Central Age 35 45 55 35 45 55

Base Rate 0.0995 0.0741 0.0328 0.2375 0.1268 0.0644

Adjustments:

Canada Adjustment 4.11% 4.11% 4.11% 3.56% 3.56% 3.56%

Definition Adjustment 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

Progressive Form Adjustment -0.0039 0.0179 0.0296 0.0024 0.0793 0.0419

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 30-Day Survival 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%


Note: The Derived Incidence Rate =

(Base Rate)

x (1 + Canada Adjustment)

x (1 + Definition Adjustment)

x (1 + 2008 Trend Adjustment)

x (1 + First-ever Incidence Adjustment)

+ (Progressive Form Adjustment)

/ (1 - Sudden Death Adjustment)

x (1 + Overlap Adjustment)

/ (1 - Prevalence Adjustment)

x (1 + 30-Day Survival Adjustment)


95

4.7 Alzheimer’s Disease

4.7.1 Definition The CLHIA Benchmark Definition is:

Alzheimer’s disease is defined as “a definite diagnosis of a progressive degenerative disease of the brain. The insured person must exhibit the loss of intellectual capacity involving impairment of memory and judgement, which results in a significant reduction in mental and social functioning, and requires a minimum of eight hours of daily supervision. The diagnosis of Alzheimer’s disease must be made by a specialist.

Exclusion: no benefit will be payable under this condition for all other dementing organic brain disorders and psychiatric illnesses.”

4.7.2 Data Sources 1. Ziegler-Graham et al. “Worldwide Variation in the Doubling Time of Alzheimer’s

Disease Incidence Rates”. Alzheimer’s & Dementia Vol. 4 Issue 5 (September 2008). Subsequently referred to as “the A&D study”.

2. German Alzheimer Society. Die Epidemiologie der Demenz (The Epidemiology of Dementia). June 2008 Fact Sheet 1 (www.deutsche-alzheimer.de/fileadmin/alz/pdf/factsheets/FactSheet01.pdf). Subsequently referred to as “the German study”.

0,0

0,1

0,1

0,2

0,2

0,3

0,3

15

19

23

27

31

35

39

43

47

51

55

59

63

67

71

75

79

83

Rat

e pe

r 1,0

00

Age Last

Multiple Sclerosis Incidence Rates Raw vs Derived


http://www.deutsche-alzheimer.de/fileadmin/alz/pdf/factsheets/FactSheet01.pdf�

http://www.deutsche-alzheimer.de/fileadmin/alz/pdf/factsheets/FactSheet01.pdf�


96

3. Gauthier, S. “Advances in the pharmacotherapy of Alzheimer's disease”. Canadian Medical Association Journal March 5, 2002 (http://www.cmaj.ca/cgi/content/full/166/5/616?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&titleabstract=dementia&searchid=1029696482444_999&stored_search=&FIRSTINDEX=10&journalcode=cmaj). Subsequently referred to as “the CMAJ review” for short.

4. Johnson et al. “Modeling the Effect of Alzheimer’s Disease on Mortality”. The International Journal of Biostatistics Vol. 3 Issue 1 (2007) (www.bepress.com/ijb/vol3/iss1/13). Subsequently referred to as “the IJB study”.

5. Kokmen et al. “Is the Incidence of Dementing Illness Changing? A 25-year Time Trend Study in Rochester, Minnesota (1960–1984)”. Neurology 43 (1993): 1887. Subsequently referred to as “the Rochester study”.

4.7.3 Calculation of Base Incidence Rates We followed the general approach set out in section 3.4. We address points specific to the development of incidence rates for Alzheimer’s disease below.

Raw Observed Incidence Rates For ages 65–90, we used the incidence rates from the A&D study as the primary basis. This study is a comprehensive summary of existing Alzheimer’s disease research. It identifies all studies in the peer review literature that reported age-specific incidence rates for Alzheimer’s disease and models the logarithm of the incidence rates as a linear function of age. Twenty-eight articles were identified (12 of them North American) and two of them were excluded (both of them North American) because either a more recent study with updated data was identified or only one incidence rate for a single age group was provided.

To determine if the worldwide data are representative of Canadian experience, we used the same methodology as the A&D study to fit a curve to the North American data points only. Note that we included the North American data points that were originally excluded since we wanted to incorporate as much information as possible. Also, the A&D study focused on the doubling time, i.e., slope of the curve, while we are looking for the best estimate incidence rates.

The following graph shows observed and fitted incidence rates of Alzheimer’s disease as a function of age from the A&D study for all worldwide data and only North American data points (in red).

Worldwide fitted incidence rates were calculated as Incidence (age) = 0.117 exp{0.1271(age-60)}.

North American fitted incidence rates were calculated as Incidence (age) = 0.141 exp{0.1281(age-60)}.

The dashed lines represent upper and lower bounds on the predicted incidence.

http://www.cmaj.ca/cgi/content/full/166/5/616?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&titleabstract=dementia&searchid=1029696482444_999&stored_search=&FIRSTINDEX=10&journalcode=cmaj�



http://www.bepress.com/ijb/vol3/iss1/13�


97

While the slopes look almost identical, North American incidence rates appear to be more than 20% higher than worldwide. Hence we used the North American curve to estimate expected Canadian experience.

Next, we compared the estimates for the A&D study and the German study. While both sources combined results of many worldwide studies, the German study did not mention what studies their rates were based on. Still, it is good to note that the Alzheimer’s disease incidence rates (estimated as two-thirds of all dementia) were quite similar to the North American rates from the A&D study. Note that the Rochester study, which is one of the studies included in the North American data of the A&D study, also finds that about two-thirds of dementia cases involved Alzheimer’s disease (67.6% of clinical cases and 65.1% of all cases).

Here is a summary of the difference rates:

The German study also reported that there was no material variation in incidence rates by gender, which we confirmed by comparing male and female data points in the A&D study. Hence the same Alzheimer’s disease incidence curve was assumed for males and females.

Worldwide Incidence Rates (expressed as % per year)

Dementia Alzheimer's Disease Only(1) (2) (3)

Published (Worldwide) North American Only Age Group Midpoint =0.117*EXP(0.1271*(age-60)) =0.141*EXP(0.1281*(age-60)) (2/3 of total dementia)

65-69 67 0.28 0.35 121% 0.4 0.27 130%70-74 72 0.54 0.66 122% 0.9 0.60 109%75-79 77 1.02 1.24 123% 1.9 1.27 98%80-84 82 1.92 2.36 123% 4.1 2.73 86%85-89 87 3.62 4.48 124% 6.5 4.33 103%90+ 90 5.30 6.58 124% 10.1 6.73 102%

(2) / (1)

Alzheimer's & Dementia(Alzheimer's Disease)

German Alzheimer's Society

(2) / (3)


98

Note that the German study explains the higher prevalence of Alzheimer’s disease among women by their higher life expectancy.

The graph above also shows that while there is a good fit of the curve under age 90, incidence rates seem to level off at this age. Hence fitted incidence rates have been capped at age 90 rates.

While few credible data exist for ages below 65, we assumed that the same exponential growth that was observed above age 65, applies under age 65. Hence the same formula was used for ages below 65 as for ages 65 to 90. As expected, this led to very low incidence rates for younger ages and provided a reasonable fit to the few data points available below 65.

Definition Adjustment A number of the studies included in the A&D study used the MMSE score in their definition of Alzheimer’s disease. The following table from the CMAJ review shows that it takes two to three years to deteriorate from mild to moderate Alzheimer’s disease requiring supervision.

Since the benchmark definition most closely resembles the moderate stage, a two-year parallel shift was assumed to adjust the raw observed incidence rates to the benchmark definition. Alzheimer’s disease is a degenerative disease that progresses over a number of years and does not normally lead to death in the early stages of the disease. The IJB study, for example, finds that “Generally, death from Alzheimer’s disease occurs only once patients have progressed to the late stage of disease” and that “Persons in the early stage of disease have mortality rates similar to persons without disease”. Hence we assumed that a two-year survival adjustment during the period of the shift was not needed.

4.7.4 Trends We found no material variation in incidence rates between earlier and later data points in the A&D study. In addition, the German study comments on the fact that there is no evidence that the risk of dementia has changed over the last decades. Hence no trend adjustment was made.

4.7.5 First-ever Adjustment We made no adjustment since the incidence rates reported were already based on Alzheimer’s-free samples.


99

4.7.6 Sudden Death Adjustment Since Alzheimer’s disease is a degenerative disease that progresses over a number of years, no sudden death adjustment was made.

4.7.7 Overlap Adjustment The main overlap is between Alzheimer’s Disease and Loss of Independent Existence (LOIE). Because not all CI products offer coverage for LOIE and because it can be available as a rider, the subcommittee decided that LOIE would come last in line for the overlap adjustments, i.e., that Alzheimer’s disease leading to LOIE would be removed from LOIE, not from Alzheimer’s Disease (please refer to section 4.9.7 for the details). Hence we made no overlap adjustment here.

4.7.8 Prevalence Adjustment No prevalence adjustment is required since the incidence rates were developed based on Alzheimer’s-free samples.

4.7.9 Thirty-day Survival Adjustment Since Alzheimer’s disease is a degenerative disease that progresses over a number of years, it does not normally lead to death in the early stage of the disease. The IJB study, for example, found that Alzheimer’s disease “increase[s] background mortality rates by about 8% per year [additive constant regardless of age or gender] once a patient progresses to the late stage of disease.” Since we used a two-year age adjustment to model the onset of the moderate stage of Alzheimer’s disease, we have assumed that there is no increase in the background mortality rate during the first 30 days. Hence we adjusted the incidence rates to reflect standard mortality (please refer to section 3.8 for the details.)


100


A comparison of the raw base rates and derived incidence rates per 1,000 for ages 15–90 is graphed below.


Adjustments:Definition Adjustment -22.6% -22.6% -22.6% -22.6% -22.6% -22.6%2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%First Ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%Prevalence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%30-Day Survival 0.0% -0.2% -1.2% 0.0% -0.1% -1.0%


Note: The Derived Incidence Rate = (Base Rate)x (1 + Definition Adjustment)x (1 + 2008 Trend Adjustment)x (1 + First Ever Incidence Adjustment)/ (1 - Sudden Death Adjustment)x (1 + Overlap Adjustment)/ (1 - Prevalence Adjustment)x (1 + 30-Day Survival Adjustment)

Summary of Calculation of Incidence Rates per 1,000: Alzheimer's DiseaseMales Females


101

4.8 Parkinson’s Disease 4.8.1 Definition The CLHIA Benchmark Definition is:

Parkinson’s Disease means primary idiopathic Parkinson’s disease, which is characterized by a minimum of two or more of the following clinical manifestations:

• Muscle rigidity; • Tremor; and • Bradykinesis (abnormal slowness of movement, sluggishness of physical and

mental responses).

The insured must require substantial physical assistance from another adult to perform at least two of the following activities of daily living (ADL):

• Bathing—the ability to wash oneself in a bathtub, shower or by sponge bath, with or without the aid of equipment;

• Dressing—the ability to put on, remove, fasten and unfasten all necessary clothing, braces, artificial limbs or other surgical appliances;

• Toileting—the ability to get to and from the toilet and complete related personal hygiene;

• Bladder and bowel continence—the ability to manage bowel and bladder functions, with or without any protective undergarments or surgical appliances, so that a reasonable level of hygiene is maintained;

Alzheimer's Disease Base vs Derived Incidence Rates

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90

Age Last

Rate

per

1,0

00

Base Male Base Female Derived Male Derived Female


102

• Transferring—the ability to move into and out of a bed, chair or wheelchair, with or without the use of equipment; and

• Feeding—the ability to consume food that has already been prepared and made available, with or without the use of adaptive utensils.

Exclusion: no benefit will be payable under this critical condition for all other types of Parkinsonism.

4.8.2 Data Sources 1. Driver et al. “Incidence and Remaining Lifetime Risk of Parkinson’s Disease in

Advanced Age”. Neurology 72 (2009): 432–438. Subsequently referred to as “the Physicians’ Health Study”.

2. Mehta et al. “Population Prevalence and Incidence of Parkinson’s Disease in an Australian Community”. Internal Medicine Journal, June 2007. Subsequently referred to as “the Australia study”.

3. Kontakos, N., and J. Stokes. “XII. Parkinson’s Disease—Recent Developments and New Directions”. Part of a monograph series on aging-related diseases in Chronic Diseases in Canada Vol. 20 No. 2 (2000). Subsequently referred to as “the Canada study”.

4. Canadian Institute for Health Information. Hospital Morbidity Database—Tabular Reports, 1997/1998 to 2000/2001. Subsequently referred to as “Cdn Hosp 1997 to 2001”.

5. De Pedro-Cuesta, J., and L. Stawiarz. “Parkinson’s Disease Incidence: Magnitude, Comparability, Time Trends”. Acta Neurologica Scandinavica 84 (1991): 382–388.

6. Zhang, Z.-X., and G. Román. “Worldwide Occurrence of Parkinson’s Disease: An Updated Review”. Neuroepidemiology 12 (1993): 195–208.

7. Van Den Eeden et al. “Incidence of Parkinson’s Disease: Variation by Age, Gender and Race/Ethnicity”. American Journal of Epidemiology 157 (2003): 1015–1022. Subsequently referred to as “Kaiser-Permanente study”.

8. Mayeux et al. “An Estimate of the Incidence of Dementia in Idiopathic Parkinson’s Disease”. Neurology 40 (1990): 1513.

4.8.3 Calculation of Incidence Rates Raw Observed Incidence Rate There have been several studies of Parkinson’s disease (PD) over the past two decades. However, there are limited data and very few studies describing the incidence and prevalence of Parkinson’s disease in Canada. Furthermore, as pointed out in the Canada study, “Since most patients do not require hospital care on an in-patient basis, hospital separation rates underestimate the prevalence of PD. The separation rates are also problematic in that they are not based on the number of individuals but rather on the number of discharges.”

Therefore, we chose to compile our results based on the Physicians’ Health Study. The Physicians’ Health Study method offers the following relative advantages:

• Statistics from international studies have established that the incidence of Parkinson’s disease varies greatly across different ethnic groups, socio-economic backgrounds, and environmental factors. The Physicians’ Health Study was conducted in the U.S.


103

for a primarily Caucasian population during the period 1982–2006. It is a more reasonable proxy for the Parkinson’s disease insurance risk of the Canadian case than other studies.

• Participants in the Physicians’ Health Study self-reported Parkinson’s disease on yearly follow-up questionnaires, and the ones who reported a new diagnosis of Parkinson’s disease are identified. This process effectively avoided the duplication which may appear in a hospital admission count method used in the Canada study.

• The population size is large and the follow-up period is long: a prospective cohort of 21,970 U.S. male physicians aged 40–84 years at baseline was followed up for 23 years.

• At the study entry, pre-existing conditions were screened but not to a fully underwritten extent: only very basic health checks were performed on the cohort (no Parkinson’s disease, cancer, heart attack, stroke or transient cerebral ischemia), so likely the incidence risk of this cohort still resembled that of the population basis.

• Validation test results are reported as 90% accurate: to evaluate the accuracy of the physician’s self-report of Parkinson’s disease, the study performed a validation study using 73 samples among 563 medical records which indicated a new case of Parkinson’s disease, and 66 (90%) were accurate. Among the remaining seven cases, two were not Parkinson’s disease, and five could not rule out a secondary cause of the apparent Parkinsonism. The Physicians’ Health Study retained all of the cases in deriving the raw incidence rate as can be seen from the table below, and this is consistent with the conservatism required in pricing or valuing insurance risk. Therefore, there is no need to remove these cases and we should accept the raw incidence rates as stated in the Physicians’ Health Study.

The raw observed incidence rates are based on table 1, page 433 of the Physicians’ Health Study:

Age-specific and Overall Male Annual Incidence Rates of Parkinson’s Disease per 100,000 Person-years

Age Group No. of Cases Person-years Incidence Rate 40–44 0 12,553.0 0 45–49 2 34,324.0 5.83 50–54 9 54,448.5 16.53 55–59 28 72,172.0 38.80 60–64 57 84,889.0 67.15 65–69 101 75,356.5 134.03 70–74 111 58,443.0 189.93 75–79 101 39,660.0 254.67 80–84 80 22,512.5 355.36 85–89 59 9,597.5 614.74 90–99 15 3,361.0 446.30 Total 563 467,316.5 120.48

Source: table 1, page 433, the Physicians’ Health Study.

The crude annual incidence rate of Parkinson’s disease in the study population was 120.48 cases/100,000 person-years. The age-specific incidence rate increased sharply beginning at


104

age 60, peaked in age 85 to 89, and declined after age 90. However, the Physicians’ Health Study made no conclusion regarding to the incidence rate pattern for individual age 90 or older because “it is unclear whether the observed decline represents a true decrease in the risk of Parkinson’s disease or is simply due to under-diagnosis”. Thus we only included age groups from 40 to 89 in deriving the raw incidence rate.

The Physicians’ Health Study did not include individuals younger than 40 years of age. Therefore, we needed to secure a second source if we were to establish incidence rates for ages 20 to 39. The Australian study derived a basically flat incidence rate of 1 per 100,000 for males aged from 20 to 38, then increased to 5.2 per 100,000 at age 45. This rate is reasonably close to 5.83 in the Physicians’ Health Study; therefore, we combined these together as the raw observed incidence rates.

Adjustment for Parkinson’s Disease Incidence Rates Due to the Difference in Definitions The definition of Parkinson’s disease in the Physicians’ Health Study above was based on a clinical diagnosis. Specifically, a Parkinson’s disease case was considered valid if a review of the patient’s medical records revealed one or more of the following:

1. Established diagnosis of Parkinson’s disease in the medical record or Parkinson’s disease as cause of death on the death certificate;

2. Current use of Parkinson’s disease medication such as DOPA or a DOPA agonist; 3. Neurological examination with physical findings consistent with Parkinsonism (at

least two of the following: rest tremor, rigidity, bradykinesia or postural instability) with no evidence of a secondary cause;

4. Patient followed-up by a neurologist or a movement disorders specialist for Parkinson’s disease.

Male Parkinson Disease Incidence Rates

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89

Age Band

Rat

e pe

r 1,0

00

Raw Observed Rate: CombinedAustralia (20-44) and PhysicianStudy (45+)


105

Comparing to the benchmark definition in section 1, the clinic diagnostic definition in the Physicians’ Health Study was more liberal. Given the fact that no severity was included in the clinic diagnostic definition of Parkinson’s disease while ADL is used in the benchmark definition, it was necessary to adjust the raw observed incidence rates to account for the severity requirement for a valid critical illness claim purpose. A 60-month parallel shift was assumed to adjust the raw observed incidence rate from the Physicians’ Health Study.

4.8.4 Adjustment for Trend The Physicians’ Health Study had compared its research result to that of the Rochester Epidemiology Project (1935–1979); the cumulative incidence of Parkinson’s disease from age 45 to 100 was remarkably similar although the lifetime risk3

More explicit evidence may be found in Zhen-Xin Zhang and Gustavo C. Roman’s paper “Worldwide Occurrence of Parkinson’s Disease: An Updated Review”. In this study, the

was quite different. This is because the increased longevity had an impact on lifetime risk but not on the raw incidence rate. The comparison between the Physicians’ Health Study and the Rochester Epidemiology Project indicated that a trend may exist in the lifetime risk measure due to an increasing life span but not in the raw incidence rate.

3 Lifetime risk estimates the absolute risk of developing Parkinson’s disease before dying of some other cause. Neurology 72 (2009): 433—the Physicians’ Health Study.

Male Parkinson’s Disease Incidence Rates

0.0

0.5 1.0

1.5 2.0

2.5 3.0

3.5 4.0

4.5 5.0

5.5 6.0

6.5

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 Age Band

Raw Observed Rate: Combined Australia (20-44) and Physician Study (45+) Adjusted for Benchmark Definition

Rate per 1,000


106

researchers reviewed Parkinson’s disease surveys over 27 regional populations and analyzed Parkinson’s disease frequency from 45 communities. They concluded that “with regard to time trends, there were no substantial changes in Rochester., Minn, from 1955 to 1965. Similarly, crude incidence in Sardinia from the years 1961-1971, and in Iceland from 1954 to 1963, and age-adjusted incidence in Rochester, from 1945 through 1979 showed no marked temporal fluctuation.”

In short, there appeared to be no clear trend associated with the Parkinson’s disease incidence rate over time; therefore no adjustment was made to account for any trend.

4.8.5 Adjustment for First-ever Incidence The Physicians’ Health Study’s population cohort was fixed and has been individually identified. All Parkinson’s disease cases were self-reported and this design explicitly excluded the possibility of individuals reporting their same Parkinson’s disease case at a different site or different time. Given the fact that an individual diagnosed with Parkinson’s disease does not recover and contract a second instance of PD, the incidence rate derived was already on a first-ever basis and there is no need for adjustment.

4.8.6 Adjustment for Sudden Deaths PD is a chronic disease, and the deterioration of the patient’s health is slow relative to other critical illnesses. Therefore, no adjustment was made for sudden deaths.

4.8.7 Adjustment for Overlap There is overlap to consider between the critical conditions of PD, Alzheimer's and Loss of Independent Existence (dementia can satisfy cognitive impairment for LOIE). The article “An Estimate of the Incidence of Dementia in Idiopathic Parkinson’s Disease” examines the link between dementia and PD. None of the major studies had established any relationship between incidence of Parkinson’s disease and the existence of the other critical illness covered conditions.

The Physicians’ Health Study, which was used as the general incidence basis for PD, had excluded reported cases of Parkinson’s disease in patients who developed dementia within the first year of Parkinson’s disease diagnosis. Further, based on the diagnostic criteria used in the Physicians’ Health Study, we are confident that a Parkinson’s disease diagnosis would not occur after a diagnosis of Alzheimer's. Therefore, we are comfortable that the developed Parkinson’s disease incidence rates are “first CI” in the overlap group of PD, Alzheimer’s and LOIE.

Incidence rates for Alzheimer’s and LOIE need to consider potential overlap with Parkinson’s disease in their respective developments.

4.8.8 Adjustment for Prevalence The study population has been cleared of any pre-existing Parkinson’s disease and other major health issues at study entry and thus no adjustment for prevalence was deemed necessary.

4.8.9 Adjustment for the 30-day Survival Period As mentioned in section 4.8.6, Parkinson’s disease progresses slowly and does not generally lead to death within the 30-day survival period required by critical illness insurance. Therefore, we adjusted the incidence rates as described in section 3.8.


107

4.8.10 Summary of Rates and Adjustments Summary of Calculation of Incidence Rates per 1,000: Parkinson’s Disease Male

Central Age 42 47 52 57 62 67 72 77 82 87 Raw Observed Rate

0.027 0.0583 0.1653 0.3880 0.6715 1.3403 1.8993 2.5467 3.5536 6.1474

Adjustment: Diff. in Definitions -0.017 -0.031 -0.107 -0.223 -0.284 -0.669 -0.559 -0.647 -1.007 -2.594

2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 30-Day Survival (%) -0.01%

-1.12%

-0.03% 0.13% 0.05%

-0.11%

-0.22%

-0.38%

-0.57%

-0.90%

Derived Incidence Rate 0.0100

0.0270

0.0583

0.1652

0.3877

0.6706

1.3373

1.8925

2.5321

3.5214

Comparisons of the raw and derived incidence rates as well as a Lagrange fitted curve between the central points are graphed below:

Male Parkinson's Disease Incidence Rates

0.0

0.51.0

1.52.0

2.5

3.03.5

4.0

4.55.0

5.56.0

6.5

7.07.5

8.0

20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94

Age

Rat

e pe

r 1,0

00

Male Raw Observed Rate: CombinedAustralia (20-44) and Physician Study (45+)

Male Adjusted for 30-day Survival Period

Male Central Age Points

Male Lagrange fitted curve


108

4.8.11 Adjustment for Female The study results on the incidence rate of Parkinson’s Disease among females are mixed. Although the Physicians’ Health Study did not address the incidence of Parkinson’s Disease in women directly, it estimates that “women have a twofold decreased risk of Parkinson’s Disease compared with men, which seems to persist even to very old ages.” Similarly, research from the Kaiser-Permanente study (a California HMO), suggested that the rate for men is 91% higher than that for women. On the other hand, the Australian study concludes that Parkinson’s Disease appears to have only a slightly higher incidence rate among males than females, and it applied a factor of 0.85 for female in respect of male rate. In their paper The Worldwide Occurrence of Parkinson’s Disease: An Updated Review, Zhen-Xin Zhang and Gustavo C. Roman observed “after age standardizations, in all but one study, the male over female ratios were close to one or slightly greater in men than in women”.

In aggregate, the studies suggest that incidence is higher in males than females. Based on the incidence rate data from the Kaiser-Permanente study, the female ratios tend to decline with increasing age. This general trend is also observed in the Canadian hospital separation data. While hospital separations cannot be readily used to develop first-ever incidence rates, they can provide additional insights into male: female ratios. The following considerations were noted in the development of the recommended factors:

• There is limited incidence of Parkinson’s disease under the age of 50. • Kaiser-Permanente study data above age 80 were limited—under 100 cases of

diagnosed Parkinson’s disease. • Applying a floor of 55% for the female relative factors provides some conservatism,

which is prudent due to the mixed reports in the various academic research publications.

The “recommended” column in the above table was used to derive the Parkinson’s disease incidence rate for females from that of males, and after the 30-day survival period adjustment, the rates summary and graph are as below:

Development of Incidence Relative Factors for Females (as % Males)

Age K-P StudyCdn Hosp 1991-1995

Cdn Hosp 1997-2001 Recommended

44 & Under 47.1% 78.5% 63.6% 80%45-49 47.1% 78.5% 63.6% 80%50-54 77.5% 78.5% 63.6% 80%55-59 77.5% 78.5% 63.6% 75%60-64 58.6% 78.5% 63.6% 70%65-69 58.6% 74.5% 69.3% 65%70-74 55.7% 61.2% 69.3% 60%75-79 55.7% 59.7% 48.4% 55%80-84 37.1% 50.4% 48.4% 55%85 & over 37.1% 38.0% 48.4% 55%


109

4.8.12 Summary of Rates and Adjustments Summary of Calculation of Incidence Rates per 1,000: Parkinson’s Disease

Female Central Age 42 47 52 57 62 67 72 77 82 87 Male Raw Observed Rate

0.027 0.0583 0.1653 0.3880 0.6715 1.3403 1.8993 2.5467 3.5536 6.1474

Gender Adj: Diff. in Gender 80% 80% 80% 75% 70% 65% 60% 55% 55% 55%

Est. Female Raw Rate 0.0216 0.0466 0.1322 0.2910 0.4701 0.8712 1.1396 1.4007 1.9545 3.3811

Other Adj: Diff. in Definitions -0.0136 -0.0250 -0.0856 -0.1588 -0.1791 -0.4011 -0.2684 -0.2611 -0.5538 -1.4266

2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0%

Overlap 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Prevalence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 30-Day Survival (%) -0.01% -0.01% 0.06% 0.00% -0.05% -0.10% -0.15% -0.25% -0.42% -0.72% Derived Incidence Rate 0.0080 0.0216 0.0466 0.1322 0.2908 0.4696 0.8699 1.1368 1.3948 1.9404

Comparisons of the male raw and female derived incidence rates as well as a female Lagrange fitted curve between the central points are graphed below:


110

4.9 Loss of Independent Existence

4.9.1 Definition The CLHIA Benchmark Definition is:

Loss of independent existence is defined as “a definite diagnosis of:

a) a total inability to perform, by oneself, at least two of the following six activities of daily living, or,

b) Cognitive impairment, as defined below,

for a continuous period of at least 90 days with no reasonable chance of recovery. The diagnosis of loss of independent existence must be made by a specialist”.

Activities of daily living are:

• Bathing—the ability to wash oneself in a bathtub, shower or by sponge bath, with or without the aid of equipment;

• Dressing—the ability to put on and remove necessary clothing including braces, artificial limbs or other surgical appliances;

• Toileting—the ability to get on and off the toilet and maintain personal hygiene;

• Bladder and bowel continence—the ability to manage bowel and bladder function with or without protective undergarments or surgical appliances so that a reasonable level of hygiene is maintained;

Female Parkinson's Disease Incidence Rates

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94

Age

Rat

e pe

r 1,0

00

Male Raw Observed Rate: CombinedAustralia (20-44) and Physician Study(45+)Male Adjusted for 30-day Survival Period

Female Adjusted for 30-day SurvivalPeriod

Female Lagrange fitted curve


111

• Transferring—the ability to move in and out of a bed, chair or wheelchair, with or without the use of equipment; and

• Feeding—the ability to consume food or drink that already has been prepared and made available, with or without the use of adaptive utensils.

Cognitive impairment is defined as:

“Mental deterioration and loss of intellectual ability, evidenced by deterioration in memory, orientation and reasoning, which are measurable and result from demonstrable organic cause as diagnosed by a specialist. The degree of cognitive impairment must be sufficiently severe as to require a minimum of eight hours of daily supervision.”

Determination of a cognitive impairment will be made on the basis of clinical data and valid standardized measures of such impairments.”

Exclusion: no benefit will be payable under this condition for any mental or nervous disorder without a demonstrable organic cause.

4.9.2 Data Sources 1. Society of Actuaries. Intercompany Study 1984–2004. Long-term Care Experience

Committee, November 2007.

2. Society of Actuaries. 1985 National Nursing Home Survey Utilization Data. Report of the Long-term Care Experience Committee, 1988–1990 reports. Transactions of Society of Actuaries.

3. Society of Actuaries. Non-insured Home and Community-based Long-term Care Incidence and Continuance Tables. Non-insured Home and Community Experience Subcommittee of the Long-term Care Experience Committee. October 2, 1999.

4. Gilmour, H., and J. Park. Dependency, Chronic Conditions and Pain in Seniors Supplement to Health Reports, Vol. 16. Statistics Canada.

5. National Institute on Aging. The Health & Retirement Study: Growing Older in America. U.S. National Institutes of Health, 2007.

4.9.3 Calculation of Incidence Rates Raw Observed Incidence Rates We used as a starting point the results of the SOA LTCI intercompany study for the 2002–2004 observation period (data source 1). The definition for LTC insurance is similar to our LOIE definition, except for the “no reasonable chance of recovery” part. The rates in the study are sex-distinct and grouped in age bands. Different elimination periods are available in that study, and we used the 90-day elimination period, in line with the 90-day requirement in the LOIE definition. The incidence rates are:


112

Incidence Rates from SOA Intercompany LTCI study, 90-day Elimination Period, U.S., 2002-2004

Age Male Female < 40 0.009% 0.007%

40–49 0.019% 0.019% 50–59 0.036% 0.038% 60–64 0.089% 0.082% 65–69 0.123% 0.149% 70–74 0.287% 0.382% 75–79 0.757% 1.014% 80–84 1.835% 2.440% 85–89 3.963% 5.152% 90+ 8.192% 9.304%

We then interpolated/extrapolated these raw incidence rates over the full age range using a three-point Lagrange interpolation formula. For ages 18–36, age 37 rates have been used as a floor.

Two adjustments were necessary to convert to Canadian population rates. First we adjusted the rates to convert the U.S. rates to Canada rates, using the results of the Statistics Canada report Dependency, Chronic Conditions and Pain in Seniors (data source 4) on prevalence of ADL (activities of daily living) dependency in 2003, and the results of the National Institute on Aging study in the U.S. in 2002 (data source 5). Both of these studies show the percentage of seniors requiring help with one ADL, so this allows us to compare Canadian experience to U.S. experience. Canadian experience is lower, as shown below:

Canada U.S. 65–74 75–84 85+ 65–74 75–84 85+ Men 4.0% 8.0% 20.0% 4.5% 8.0% 16.2% Women 4.0% 9.0% 23.0% 5.2% 10.7% 27.8%

Using this information and 2001 census data to calculate an overall prevalence rate, the adjustment from U.S. to Canada is 95.1% for male and 80.3% for female. For conservatism, and because the two studies are not identical and may have had slightly different methodologies and definitions, we used 95% for both males and females.

The second adjustment we made is to convert to general population, given that the SOA LTCI intercompany study results are for an insured population. For that adjustment, we used the SOA 1985 nursing home survey (data source 2). When the rates of that survey were developed, two categories were created: one for all stays and one for insurable stays, i.e., nursing home stays for individuals that would not have qualified for insurance. We used the difference between those two set of rates to estimate the difference of insured population versus general population. The result is 6.6% for males and 5.2% for females.


113

The resulting raw LOIE rates are:

Age Male Female < 40 0.01% 0.01%

40–49 0.02% 0.02% 50–59 0.04% 0.03% 60–64 0.09% 0.07% 65–69 0.12% 0.13% 70–74 0.29% 0.32% 75–79 0.77% 0.86% 80–84 1.86% 2.06% 85–89 4.02% 4.35% 90+ 8.30% 7.86%

Adjustment for Definition The next step was to adjust the definition to exclude the LTC claims that went back to an active status, given that the LOIE definition is for cases with no reasonable chance of recovery only. To do that, we looked at the claims persistency rates compared to the mortality rate while disabled, with the difference being equal to the proportion of cases going back to active status. The results are:

Year

SOA Study Persistency

SOA Study Disabled Lives

Mortality Difference

1 64.7% 78.7% 14.0% 2 46.3% 61.9% 15.6% 3 33.1% 48.7% 15.6%

Based on those results, we decreased the incidence rates by 15% for all ages up to 75; then gradually graded them to 0 at age 90+ to reflect that, as people get older, their chances of recovering from a loss of independence are smaller.

4.9.4 Adjustment for Trend The data require adjustment for trend from 2003 (the midpoint of the study period for the base data) to January 2008.

The SOA published intercompany studies in 1999, 2001 and 2004. Ideally, we would have used the 2004 exposure on the previous years’ results by age to look at the trend. However, because of error in allocating the exposure correctly by age group for the 1999 and 2001 reports resulting in erroneous incidence rates, it was not possible. We did, however, look at the overall incidence rate (all ages combined) from 1999 to 2004, and the results for 1999, 2001 and 2004 were respectively 0.6%, 0.68% and 0.64%. The weighted average by exposure for the three studies is 0.64%, so we decided not to use a trending factor.

4.9.5 Adjustment for First-ever Incidences Given that we eliminated LTC claims that would result in a return to active status in section 4.9.3 and are using incidence rates that only lead to death, we did not further adjust for first-ever incidence.


114

4.9.6 Adjustment for Sudden Deaths There is no sudden-death adjustment required, since our base data already have a 90-day elimination period.

4.9.7 Adjustment to Remove Overlap with Other Critical Illnesses Because not all CI products offer coverage for LOIE, or because it can be available as a rider, the subcommittee decided that LOIE would come last in line for the overlap adjustments, i.e., that all other critical illness covered in this report and leading to LOIE would be removed.

Once again, we used the SOA LTCI intercompany study to determine the overlap adjustments. That report has a section on causes of total claims broken down by cause. The results are as follows:

Age All <65 65–74 75+

Alzheimer’s 24% 9% 21% 26% Stroke 12% 10% 13% 12%

Arthritis 10% 12% 10% 10% Cancer 10% 28% 17% 8% Injury* 10% 9% 9% 10%

Circulatory 10% 4% 7% 10% Nervous system and sense

organs* 6% 17% 7% 5% Respiratory 5% 2% 5% 5%

Digestive system 2% Mental 2% Other 10% 9% 11% 14%

The boxes that are highlighted correspond to conditions covered in this report and are excluded. Because Alzheimer’s include all types of dementia, we made the same adjustment that was done for Alzheimer’s earlier in this report, i.e., only exclude two-thirds of Alzheimer’s cases.

The categories with an asterisk also include conditions covered in this report like loss of limbs, severe burns, multiple sclerosis, blindness and Parkinson’s disease.

We looked at two combinations in the list above: highlighted conditions only, and highlighted conditions plus 50% of conditions with an asterisk. The results are:

Highlighted only 48% 51% 47% Highlighted with 50% of * 61% 59% 55%

Based on those results, we decided to use 60% up to age 70, grading down to 55% at ages 80-plus, to remove overlap with other critical illnesses covered.

4.9.8 Adjustment for Prevalence The study used in section 4.9.3 used to convert insurance population to general population uses the same population in the denominator, therefore an adjustment for prevalence is required.


115

As mentioned in section 4.9.3, data source 4 has statistics on prevalence of at least one ADL dependency in Canada.

Canada 65-74 75-84 85+ Men 4.0% 8.0% 20.0% Women 4.0% 9.0% 23.0%

We had to adjust that number for a two or more ADL dependency, to be in line with the LOIE definition. For that adjustment, we used the SOA Non-insured Home and Community-based Long-term Care Incidence and Continuance Tables (data source 3). Those tables are an attempt to model incidence and prevalence rates for home-based care, with different elimination periods and benefit triggers. For a 90-day elimination period, the differences in the prevalence rate are:

Age 1+ ADL 2+ ADL 65 5.44% 2.45% 70 4.36% 2.64% 75 6.35% 3.88% 80 10.59% 6.24% 85 14.65% 8.13% 90 20.97% 12.53%

Those ratios between 2+ and 1+ were applied to the rates above. We then interpolated/extrapolated prevalence rates using a three-point Lagrange interpolation formula (with 0 as the value at age 18).

4.9.9 Adjustment for the Survival Period There is no survival period adjustment required, since our base data already have a 90-day elimination period.


Summary of Calculation of Incidence Rates per 1,000: LOIE Males Females Central Age 35 55 75 35 55 75 Raw Observed Rate 0.0946 0.3692 5.3909 0.0692 0.3753 7.1248 Adjustments: LOIE excl. from Definition -15.0% -15.0% -15.0% -15.0% -15.0% -15.0% 2008 Trend 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% First-ever Incidence 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Sudden Death 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Overlap -60.0% -60.0% -57.5% -60.0% -60.0% -57.5% Prevalence 0.0% 0.1% 3.8% 0.0% 0.1% 3.8% 30-Day Survival 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% Derived Incidence Rate 0.0322 0.1256 2.0248 0.0235 0.1277 2.6760

Note: The Derived Incidence Rate for LOIE =


116

(Raw Observed Rate) x (1 + LOIE Excl. from Definition Adjustment) x (1 + 2008 Trend Adjustment) x (1 + First-ever Incidence Adjustment) / (1 - Sudden Death Adjustment) x (1 + Overlap Adjustment) / (1 - Prevalence Adjustment) x (1 + 30-Day Survival Adjustment)


4.10 Minor Conditions Overview A number of conditions are included in critical illness policies which have limited incidence rates. The development of these rates will be brief. In all cases, the CLHIA Benchmark Definition will be adopted for the purposes of rate development.

For each condition, the definition and a brief summary of the incidence rate development is provided. For all conditions, a specialist is a licensed medical practitioner who has been trained in the specific area of medicine relevant to the covered critical illness condition for which benefit is being claimed, and who has been certified by a specialty examining board. In the absence or

0.0000

20.0000

40.0000

60.0000

80.0000

100.0000

120.0000

140.0000

160.0000

180.0000

18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99

Rate

Age

LOIE population incidence rates

Raw Male

Raw Female

Derived Male

Derived Female


117

unavailability of a specialist, and as approved by the insurer, a condition may be diagnosed by a qualified medical practitioner practising in Canada or the U.S.


We did not make adjustments for prevalence, overlap, and survival due to materiality within the framework of overall incidence rates. We have used statistics from the Staple Inn Actuarial Society’s Exploring the Critical Path report for certain conditions where comparable Canadian statistics were not readily available, and where we thought country-specific differences were immaterial within the framework of overall incidence rates.

4.10.1 Bacterial Meningitis 4.10.1.1 Definition The CLHIA Benchmark Definition is:

Bacterial meningitis is defined as a definite diagnosis of meningitis, confirmed by cerebrospinal fluid showing growth of pathogenic bacteria in culture, resulting in neurological deficit documented for at least 90 days from the date of diagnosis. The diagnosis of bacterial meningitis must be made by a specialist.

4.10.1.2 Data Sources 1. Public Health Agency of Canada. Bacterial Meningitis In Canada:

Hospitalizations (1994–2001). Canada Communicable Disease Report, Vol. 31 No. 23.

2. Schuchat et al. “Bacterial Meningitis in the United States in 1995”. The New England Journal of Medicine 337 (1997): 970–976.

4.10.1.3 Incidence Rates Incidence rates were derived from U.S. statistics. Canadian data were broadly consistent with U.S. data, although incidence rates by age were not available. Survival rates were based on overall Canadian survival rates for bacterial meningitis, adjusted proportionally using Statistics Canada Canadian population mortality by age.

Incidence Rate

per 1,000 Age Male Female 20 0.016 0.016 30 0.016 0.016 40 0.016 0.016 50 0.018 0.018 60 0.020 0.020 70 0.021 0.021 80 0.014 0.014 90 0.000 0.000


118

4.10.2 Blindness

4.10.2.1 Definition The CLHIA Benchmark Definition is:

Blindness is defined as the definite diagnosis of the total and irreversible loss of vision in both eyes, evidenced by:

• The corrected visual acuity being 20/200 or less in both eyes; or • The field of vision being less than 20 degrees in both eyes.

The diagnosis of blindness must be made by a specialist.

4.10.2.3 Incidence Rates Incidence rates are interpolated based on figures from the Staple Inn Actuarial Society’s Exploring the Critical Path report.

Incidence Rate


4.10.3 Coma 4.10.3.1 Definition The CLHIA Benchmark Definition is:

Coma is defined as the definite diagnosis of a state of unconsciousness with no reaction to external stimuli or response to internal needs for a continuous period of at least ninety-six (96) hours, and for which period the Glasgow coma score must be four (4) or less. The diagnosis of coma must be made by a specialist.

Exclusions: no critical illness benefit will be payable for:

• A medically-induced coma; • A coma which results directly from alcohol or drug use; or • A diagnosis of brain death.

4.10.3.2 Incidence Rates Following the conclusion from the Staple Inn Actuarial Society’s Exploring the Critical Path report, we have set the incidence rate for these conditions to 0, based on the assumption overlap with other conditions accounts for materially all of the incident cases within the framework of overall incidence rates.


119

4.10.4 Deafness 4.10.4.1 Definition The CLHIA Benchmark Definition is:

Deafness is defined as the definite diagnosis of the total and irreversible loss of hearing in both ears, with an auditory threshold of ninety (90) decibels or greater within the speech threshold of five hundred to three thousand (500 to 3,000) hertz. The diagnosis of deafness must be made by a specialist.

4.10.4.2 Incidence Rates Incidence rates are interpolated based on figures from the Staple Inn Actuarial Society’s Exploring the Critical Path report:

Incidence Rate


4.10.5 Loss of Limbs 4.10.5.1 Definition The CLHIA Benchmark Definition is:

Loss of limbs is defined as the definite diagnosis of the complete severance of two (2) or more limbs at or above the wrist or ankle joint, as the result of an accidental injury or medically-required amputation. The diagnosis of loss of limbs must be made by a specialist.

4.10.5.2 Data Sources 1. Dillingham et al. “Limb Amputation and Limb Deficiency: Epidemiology and

Recent Trends in the United States”. Southern Medical Journal Vol. 95 Issue 8 (2002): 875–883.

2. National Center for Health Statistics. Health, United States, 2007, With Chartbook on Trends in the Health of Americans. Hyattsville, MD, 2007.

3. Centers for Disease Control and Prevention, National Center for Health Statistics. Compressed Mortality File 1999–2005. CDC WONDER Online Database, compiled from Compressed Mortality File 1999–2005 Series 20 No. 2K, 2008.

4. Statistics Canada. Canadian Vital Statistics, Death Database, CANSIM Table 102-0540 Deaths, by cause, Chapter XX: External Causes of Morbidity and Mortality (V01 to Y89), age group and sex, Canada, annual (number).


120

4.10.5.3 Incidence Rates Incidence rates were developed using U.S. source data, and adjusting based on differences in Canadian and U.S. mortality due to external cause.

Incidence Rate per

1,000 Age Male Female 20 0.001 0.000 30 0.001 0.000 40 0.001 0.000 50 0.001 0.000 60 0.001 0.000 70 0.001 0.000 80 0.001 0.000 90 0.001 0.001

4.10.6 Loss of Speech 4.10.6.1 Definition The CLHIA Benchmark Definition is:

Loss of Speech is defined as the definite diagnosis of the total and irreversible loss of the ability to speak as the result of accidental injury or disease for a period of at least one hundred and eighty (180) days. The diagnosis of loss of speech must be made by a specialist.

4.10.6.2 Incidence Rates Following the conclusion from the Staple Inn Actuarial Society’s Exploring the Critical Path report, we have set the incidence rate for these conditions to 0, based on the assumption overlap with other conditions accounts for materially all of the incident cases within the framework of overall incidence rates.

4.10.7 Motor Neuron Disease 4.10.7.1 Definition

The CLHIA Benchmark Definition is:

Motor Neuron Disease is defined as the definite diagnosis of one (1) of the following:

• Amyotrophic lateral sclerosis (ALS or Lou Gehrig disease); • Primary lateral sclerosis; • Progressive spinal muscular atrophy; • Progressive bulbar palsy; or • Pseudo bulbar palsy and limited to these conditions.

The diagnosis of motor neuron disease must be made by a specialist.


121

4.10.7.2 Data Sources 1. Chancellor et al. “Adult Onset Motor Neuron Disease: Worldwide Mortality,

Incidence and Distribution since 1950”. Journal of Neurology and Psychiatry 55 (1992): 1106–1115.

2. Bonaparte et al. “ALS Incidence in Nova Scotia over a 20-Year Period: A Prospective Study”. The Canadian Journal of Neurological Sciences 34 (2007): 69–73.

4.10.7.3 Incidence Rates Incidence rates are derived based on the two studies listed above.

The first paper, “Adult Onset Motor Neuron Disease: Worldwide Mortality, Incidence and Distribution since 1950”, provides a review of all publications in English or with English abstracts dealing with the frequency of adult onset MND published prior to the date of the paper (1992). The authors provide a summary of sex- and age-banded incidence rates from the papers that followed the most rigorous approach to incidence rate estimation. One of those studies related to a study in South Western Ontario.

Incidence Rate per 1,000 Age Male Female 20 0.000 0.000 30 0.010 0.007 40 0.021 0.015 50 0.041 0.028 60 0.064 0.043 70 0.070 0.048 80 0.036 0.025 90 0.000 0.000

4.10.8 Occupational HIV Infection 4.10.8.1 Definition The CLHIA Benchmark Definition is:

Occupational HIV infection is defined as the definite diagnosis of infection with the human immunodeficiency virus (HIV) resulting from accidental injury during the course of the insured’s normal occupation, which exposed the person to HIV-contaminated bodily fluids. The accidental injury leading to the infection must have occurred after the later of the effective date or the date of the last reinstatement of this policy.

Payment under this condition requires satisfaction of all of the following:

• The accidental injury must be reported to the insurer in writing within fourteen (14) days of its occurrence;

• A serum HIV test must be taken within fourteen (14) days of the accidental injury and the result must be negative;

• A serum HIV test must be taken between ninety (90) and one hundred eighty (180) days after the accidental injury and the result must be positive;


122

• All HIV tests must be performed by a duly licensed laboratory in Canada or the United States of America; and

• The accidental injury must have been reported, investigated and documented in accordance with current Canadian or United States of America workplace guidelines.

The diagnosis of occupational HIV infection must be made by a specialist.

Exclusions: no critical illness benefit will be payable for occupational HIV infection if:

• The insured has elected not to take any available licensed vaccine offering protection against HIV; or

• A licensed cure for HIV infection has become available prior to the accidental injury; or

• HIV infection has occurred as a result of non-accidental injury including, but not limited to, sexual transmission and intravenous (IV) drug use.

4.10.8.2 Data Sources 1. Centers for Disease Control and Prevention. Surveillance of Occupationally

Acquired HIV/AIDS in Healthcare Personnel, as of December 2006. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2007.

2. ———. Morbidity and Mortality Weekly Report, September 30, 2005/54(RR09); 1-17, Updated U.S. Public Health Service Guidelines for the Management of Occupational Exposures to HIV and Recommendations for Postexposure Prophylaxis. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, 2007.

3. Public Health Agency of Canada. HIV and AIDS in Canada. Surveillance Report to December 31, 2006. Surveillance and Risk Assessment Division, Centre for Infectious Disease Prevention and Control, Public Health Agency of Canada, 2007.

4.10.8.3 Incidence Rates A review of literature from Canada and the United States revealed no confirmed cases of occupational HIV infection have been recorded since 2000. The case counts fell as guidelines to prevent exposure and for prophylaxis following exposure were implemented. At this time the effective incidence rate is 0. This should be monitored as drug-resistant strains of HIV/AIDS emerge, and incidence rates adjusted accordingly as any new cases of occupationally acquired HIV are recorded.

4.10.9 Paralysis 4.10.9.1 Definition The CLHIA Benchmark Definition is:

Paralysis is defined as the definite diagnosis of the total loss of muscle function of two (2) or more limbs as a result of accidental injury or disease to the nerve supply of those limbs, for a period of at least ninety (90) days following the precipitating event. The diagnosis of paralysis must be made by a specialist.


123

4.10.9.2 Data Sources 1. Statistics Canada. Canadian Vital Statistics, Death Database, CANSIM Table

102-0540 Deaths, by cause, Chapter XX: External Causes of Morbidity and Mortality (V01 to Y89), age group and sex, Canada, annual (number).

2. Office for National Statistics. Mortality Statistics, Review of the Registrar General on deaths by cause, sex and age, in England and Wales, 2005.

3. Rick Hansen SCI Network. Inventory of Services: An Overview of SCI Services across Canada.

4.10.9.3 Incidence Rates We interpolated incidence rates based on figures from the Staple Inn Actuarial Society’s Exploring the Critical Path report, adjusted for differences in mortality due to external causes between Canada and the UK, and further adjusted based on aggregate Canadian spinal cord injury statistics.

Incidence Rate


4.10.10 Severe Burns 4.10.10.1 Definition The CLHIA Benchmark Definition is:

Severe burns is defined as the definite diagnosis of third degree burns over at least twenty percent (20%) of the body surface. The diagnosis of severe burns must be made by a specialist.

4.10.10.2 Data Sources 1. American Burn Association. National Burn Repository 2007 Report, Dataset

Version 4.0. 2. ———. Burn Incidence and Treatment in the U.S.: 2007 Fact Sheet. 3. Population Division, U.S. Census Bureau. Table 1: Annual Estimates of the

Population by Sex and Five-year Age Groups for the United States: April 1, 2000 to July 1, 2005 (NC-EST2005-01). Release date: May 10, 2006.

4. Centers for Disease Control and Prevention, National Center for Health Statistics. Compressed Mortality File 1999-2005. CDC WONDER Online Database, compiled from Compressed Mortality File 1999-2005 Series 20 No. 2K, 2008. Accessed at http://wonder.cdc.gov/cmf-icd10.html on July 17, 2008 3:23:10 p.m.

http://wonder.cdc.gov/cmf-icd10.html�


124

5. Statistics Canada. Canadian Vital Statistics, Death Database, CANSIM Table 102-0540 Deaths, by cause, Chapter XX: External Causes of Morbidity and Mortality (V01 to Y89), age group and sex, Canada, annual (number).

4.10.10.3 Incidence Rates Incidence rates were developed using U.S. source data, and adjusting based on differences in Canadian and U.S. mortality due to ICD codes X0-X19 (burn-related categories).

Incidence Rate


4.10.11 Minor Conditions Summary Male Incidence Rates

Incidence Rate per 1,000 at Selected Ages

Condition 20 30 40 50 60 70 80 90 Bacterial Meningitis 0.016 0.016 0.016 0.018 0.020 0.021 0.014 0.000 Blindness 0.010 0.010 0.010 0.020 0.064 0.118 0.182 0.256 Coma 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Deafness 0.005 0.005 0.005 0.010 0.026 0.045 0.067 0.091 Loss of Limbs 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 Loss of Speech 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Motor Neuron Disease 0.000 0.010 0.021 0.041 0.064 0.070 0.036 0.000 Occupational HIV Infection 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Paralysis 0.046 0.046 0.032 0.023 0.026 0.034 0.046 0.061 Severe Burns 0.003 0.003 0.003 0.002 0.002 0.001 0.001 0.000


125

Female Incidence Rates

Incidence Rate per 1,000 at Selected Ages Condition 20 30 40 50 60 70 80 90 Bacterial Meningitis 0.016 0.016 0.016 0.018 0.020 0.021 0.014 0.000 Blindness 0.010 0.010 0.010 0.020 0.068 0.129 0.204 0.291 Coma 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Deafness 0.005 0.005 0.005 0.010 0.027 0.048 0.073 0.101 Loss of Limbs 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 Loss of Speech 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Motor Neuron Disease 0.000 0.007 0.015 0.028 0.043 0.048 0.025 0.000 Occupational HIV Infection 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Paralysis 0.028 0.021 0.019 0.016 0.023 0.035 0.052 0.075 Severe Burns 0.002 0.002 0.001 0.001 0.001 0.001 0.000 0.000

5. ADDITIONAL CONSIDERATIONS This chapter addresses a variety of topics and is intended to provide additional insight into the population incidence rates derived in the earlier chapters as well as explore considerations for insured lives. In particular we will look at:

A) A comparison of the relative importance of each CI impairment compared to actual claims, as a way of validating the population-based incidence rates;

B) A comparison of the incidence rates to the UK’s CIBT02 stand-alone incidence rates developed in the 2006 UK Staple Inn Actuarial Society report, Exploring the Critical Path; and

C) The adjustments that might be necessary in order to adopt these tables for underwritten lives consistent with a voluntary, individually-underwritten, personally-owned insurance plan with insights borrowed from UK insured experience.

A) Comparison of Relative Importance of Population Incidence Rates to Insured Lives Chapter 2 contains a pie chart showing the distribution of actual CI claims paid by cause for males and females combined (figure 1).


126

Figure 1: Distribution of insured lives CI claims in Canada, 2010 Munich Re CI Survey.

One way to assess if the 2008 CANCI tables are reasonable is to compare the distribution of incidence by impairment predicted by the 2008 CANCI tables (figure 2) to the distribution of actual insured claims by cause as represented in figure 1.

While there should be broad similarities, the two distributions should differ: the 2008 CANCI tables represent incidence rates among the general population with some adjustments, while the actual claims distributions represents the incidence rates associated with a much healthier subset of the general population, namely an insured population, which has been rigorously screened for presence of significant risk factors.

To make this comparison we applied a hypothetical distribution to estimate the underlying insured exposures associated with figure 1.

Figure 2: Relative weight of each illness/claim trigger in 2008 CANCI Tables based on estimated insured demography.

68%

13%

5%

4% 3% 7%

Distribution by Cause 2010 Industry Claims

Males and Females

CANC HA CABG STK MS Other

53% 13% 3%

6%

6% 19%

Distribution by Cause 2008 CANCI Tables

Males and Females



127

In terms of the relative contribution of each critical illness, both figures are reasonably consistent from an ordinal perspective; cancer, heart attack, and “other” are the most significant.

The most striking difference between the 2008 CANCI tables and the actual claims distribution is the significance of the “other” category of illnesses. Before examining this further, it would be helpful to recalibrate the incidence rates of the “other” category of illnesses so that it accounts for about 7% of total claims (figure 3).

From this perspective, the 2008 CANCI tables produce a distribution of claims somewhat similar to the actual claims distribution. We note that cancer is somewhat understated and that heart attack and stroke are overstated on this basis of presentation.

Figure 3: Similar to figure 2, except recalibrated to have “Other” weighted same as insured lives.

We examined the “other” category of illnesses embedded in the 2008 CANCI tables by focusing on the incidence rates at age 45 (corresponds approximately to the average age of claim and to simplify the illustration). Figure 4 shows the proportion of the illnesses at age 45 and the “other” category now accounts for about 16% instead of 19%.

59% 17%

6% 7%

4% 7%


Males and Females-Recalibrated



128

Figure 4: Similar to figure 2, except represents age 45.

Figure 5 shows that when the “other” category is expanded the largest contributors by cause are Alzheimer’s disease, kidney failure, “partial benefits” (early skin and prostate cancer, DCIS, and angioplasty) and heart valve surgery, with each accounting for approximately 2.2% to 2.9% (i.e., 16% x 14%-16% x18%) of the total claim distribution at age 45.

53%

18%

3% 7%

3% 16%

Distribution of Claims 2008 CANCI Tables

Males and Females -Age 45



129

Figure 5: Composition of “Other” illnesses, age 45.

What could account for these weighting differences between insured lives and 2008 CANCI tables? As previously mentioned, the 2008 CANCI tables are derived from population data with some modification but do not reflect adjustments intended to reflect the process of self-selection, underwriting, and differences in the risk profiles between insured lives and the population in general. These and other factors likely account for the differences.

Experience in the UK appears to corroborate this hypothesis. Insured claims for conditions such as kidney failure, Alzheimer’s disease, major organ transplant and LOIE (total and permanent disability (TPD) in the UK) are considerably less than the population-based CIBT02 tables as compared to cancer and cardiovascular-based diseases (table 1).

3%

21%

8%

8%

14% 11%

15%

6%

14%

0%

Distribution by cause 2008 CANCI Tables

Males and Females "Other" Decomposition- Age 45

PKNS ALZ LOIE MOT KF PARTIAL BBT MINOR


130

Impairment group Insured / CIBT02 /ELT92* Cancer 71% Cardiovascular 38% Alzheimer’s 10% Kidney 19% Organ transplant 20% TPD 16% Accident type 19% Other neurological 66% Mortality* 47%*

Table 1: UK Insured experience relative to population incidence (CIBT02 tables). Source: A Critical Table Pricing Critical Illness in the UK on a new Insured Lives table, page 37, by GenRe, June 2007. ELT=English Life Tables.

Furthermore, when the insured to population (I/P) factors are applied to the 2008 CANCI tables (figure 6), the resulting expected distribution of impairments is very similar to the actual insured claims distribution in figure 1. This provides a measure of comfort that the 2008 CANCI tables are reasonable.

Figure 6: 2008 CANCI tables recalibrated to reflect UK relationships between insured lives incidence experience and population incidence experience.

71%

11%

3%

5% 4%

6%


Males and Females-Recalibrated to UK I/P



131

B) A Comparison of the Incidence Rates to the 2006 UK Staple Inn Actuarial Society report Exploring the Critical Path In this section the 2008 CANCI tables are compared to the most recent UK population-based CI tables, the CIBT02.

The CIBT02 tables (Critical Illness Base Tables for year 2002) were published on December 6, 2006, in a draft report presented by the Critical Illness Trend Research Group to the Staple Inn Actuarial Society. This comprehensive report contains a wealth of information and insights into trends impacting critical illness incidence rates, and concludes with the development of CIBT02 tables.

The CIBT02 tables are derived from population data using similar techniques as the 2008 CANCI TABLES (tables 2 and 3).

We have limited our comparison to cancer, heart attack, stroke, and multiple sclerosis as these are the most significant impairments that are based on disease rather than procedure (i.e., coronary artery bypass surgery, whose incidence is not only a function of underlying disease but also of the health care delivery system). Since this is a high-level reasonability check, we did not attempt to trend both tables to a common point in time since the trend has not been significant over this time period. For example, the cancer incidence trend applied to bring the underlying incidence rates used for the 2008 CANCI tables forward to 2008 was approximately 0% per annum for males and 0.3% per annum for females. For heart attack, stroke, and multiple sclerosis the trend rate projected from the date of the underlying data to 2008 was 0% per annum. Therefore back-casting the 2008 CANCI tables would not result in a material change for the purposes of this exercise.

Table 2: CIBT02 incidence rates, stand-alone CI, males.

MALES CIBT02-Stand -Aloneper 10,000

Age CancerHeart Attack Stroke

Multiple Sclerosis

20 2.40 0.16 0.64 0.10 25 3.22 0.44 0.82 0.29 30 4.21 1.23 1.13 0.61 35 5.68 3.43 1.69 0.93 40 8.43 9.07 2.70 1.12 45 13.95 19.96 4.55 1.14 50 24.93 32.97 7.68 1.02 55 46.47 47.14 12.56 0.83 60 83.52 65.80 20.16 0.61 65 135.08 88.65 31.53 0.40 70 198.64 114.07 49.66 0.24 75 270.68 145.34 78.11 0.13 80 328.74 185.65 113.47 0.07 85 377.88 228.86 153.93 0.04


132

Table 3: CIBT02 incidence rates, stand-alone CI, females.

To provide additional context, we examined mortality rates for cancer, heart attack, stroke, and multiple sclerosis between Canada and the UK (England and Wales) (tables 4, 5) to see if there was a consistent relationship between incidence rates and mortality rates in both regions.

FEMALES CIBT02- Stand-Aloneper 10,000

Age CancerHeart Attack Stroke

Multiple Sclerosis

20 2.07 0.04 0.51 0.25 25 3.75 0.11 0.73 0.73 30 6.69 0.30 1.02 1.47 35 11.04 0.78 1.49 2.16 40 17.20 1.93 2.34 2.54 45 27.63 4.02 3.68 2.48 50 44.45 6.91 5.43 2.13 55 63.63 11.53 7.86 1.66 60 83.10 19.87 12.05 1.17 65 101.83 32.70 19.94 0.73 70 126.03 49.93 34.27 0.41 75 159.39 73.53 57.29 0.21 80 183.05 105.09 90.22 0.11 85 200.65 140.30 139.54 0.05

Mortality Rates per 1000, England & Wales 2008 25-34 35-44 45-54 55-64 65-74 75-84

Malignant Neoplasms (C00-C97) M 0.084 0.246 0.971 3.514 9.075 18.832 F 0.106 0.370 1.112 2.958 6.252 11.712

Acute Myocardial Infarction (I22-I23) M 0.009 0.075 0.285 0.712 1.681 4.375 F 0.003 0.023 0.057 0.184 0.725 2.490

Cerebral Infraction (I63-I64) M 0.004 0.010 0.037 0.154 0.642 3.037 F 0.003 0.006 0.025 0.082 0.452 2.915

Multiple Sclerosis( G35) M 0.002 0.008 0.022 0.035 0.039 0.038 F 0.001 0.012 0.035 0.063 0.061 0.060

Total of above M 0.099 0.339 1.316 4.415 11.437 26.281 F 0.112 0.412 1.229 3.287 7.489 17.177


133

Table 4. Sources: Statistics Canada for number of deaths according to indicated cause of death codes, 2007 population; medical terms for cancer, heart attack, and stroke are malignant neoplasm, acute myocardial infarction, and cerebral infarction respectively.

Table 5. Sources: Office for National Statistics, www.statistics.gov.uk/statbase/Product. asp?vlnk=14409 (page no longer available).

A graphical depiction of the ratio between the 2008 CANCI tables and the CIBT02 for each of the indicated impairments is presented in figures 6 and 7. In addition, the ratio between Canadian and UK mortality for the same underlying impairments (cancer, heart attack, stroke, and multiple sclerosis) of mortality has been plotted.

Mortality Rates per 1000, Canada 2007 25-34 35-44 45-54 55-64 65-74 75-84

Malignant Neoplasms (C00-C97)M 0.069 0.237 1.033 3.478 8.958 17.656F 0.077 0.316 1.128 2.836 6.286 10.724

Acute Myocardial Infarction (I22-I23)M 0.006 0.061 0.248 0.737 1.743 4.177F 0.001 0.014 0.056 0.199 0.695 2.378

Cerebral Infraction (I63-I64)M 0.001 0.006 0.027 0.099 0.476 2.009F 0.000 0.004 0.013 0.053 0.325 1.689

Multiple Sclerosis (G35)M 0.000 0.004 0.012 0.023 0.047 0.035F 0.000 0.006 0.027 0.045 0.044 0.067

Total of aboveM 0.077 0.307 1.320 4.337 11.224 23.877F 0.079 0.341 1.223 3.133 7.349 14.859


134

Figure 6: Ratio of Canadian to UK incidence and mortality, males. Total represents ratio of Canadian to UK incidence for the listed impairments.

0%

20%

40%

60%

80%

100%

120%

140%

160%

20 25 30 35 40 45 50 55 60 65 70 75 80 85

Cana

dian

vs U

K (%

) Canadian vs UK Incidence, Mortality

Males

Heart Attack

Stroke

Cancer

Multiple Sclerosis

Total

Mortality


135

Figure 7: Ratio of Canadian to UK incidence and mortality, females. Total represents ratio of Canadian to UK incidence for the listed impairments.

It is reassuring to see a similar relationship between Canadian and UK total mortality and incidence. This provides our second measure of comfort that the 2008 CANCI tables are reasonable.

However, there are some interesting variations by impairment. For example, male Canadian cancer incidence is considerably higher than UK cancer incidence over ages 40–75. Based on a cursory review of detailed UK and Canadian cancer incidence data, it appears that this is primarily due to higher Canadian prostate cancer incidence.

In general, differences in population incidence and mortality rates associated with any particular impairment between two countries could be due to several factors: type and quality of health care delivery systems (impacts diagnosis and subsequent mortality), environmental factors, and other socio-economic factors. While it would be interesting to investigate these factors to help us understand all the differences, that would be beyond the scope of this paper.

C) Adjustments for Insured Lives The Living Benefits Subcommittee had originally intended to produce an experience study to measure actual individual lives experience relative to these tables and to release this research paper and the experience study concurrently. However, practical difficulties in the collection of actual experience have forced us release this research paper first.

Although the experience data covering calendar years 2002–2007 would have been very helpful in calibrating the tables for insured experience, the calibration would not have been complete since there is a paucity of claims over age 60 and over policy year seven. At best, the available experience would be helpful in calibrating early duration experience. It might also provide some important early clues with respect to the effects of smoking habits on incidence rates and the

0%

50%

100%

150%

200%

250%

300%

20 25 30 35 40 45 50 55 60 65 70 75 80 85

Cana

dian

vs U

K (%

) Canada vs UK Incidence, Mortality

Females

Heart Attack

Stroke

Cancer

Multiple Sclerosis

Total

Mortality


136

impact of policy design features such as return of premium on surrender/maturity benefits on incidence rates.

Given that the majority of individual Canadian products extend coverage to at least age 75 and given that insufficient Canadian insured experience exists, it is clear that the actuary must use considerable judgment in adjusting these tables for insurance purposes.

Listed below are some of the adjustments that should be considered to adjust population-based incidence rates for insurance costing purposes, followed by a brief discussion of each. There is considerable uncertainty and even ambiguity surrounding such adjustments. (This discussion will also briefly touch on UK CI experience where potentially relevant.)

a) Trend Trend adjustments refer to directional adjustments which bring the source incidence rates into the present, as well as the directional adjustments to project future expected levels of incidence. Many of the critical illness incidence rates derived in chapter 4 exhibited historical trends and these were extended to bring the incidence rates forward to 2008.

A more challenging problem is the projection of incidence rates into the future, because the forces leading to change in incidence appear to be influenced by known and unknown human interventions and environmental factors (e.g., excessive exposure to the sun) which could vary by impairment, gender, age, socio-economic and risk classes.

General examples of human interventions include the development of improved diagnostic technologies, pharmaceuticals, public health campaigns designed to reduce the prevalence of risk factors, and possibly, the inadvertent introduction of new or increased levels of risk factors (e.g., toxins) into the physical environment.

Another challenge is the application of aggregate-level trends to segmented-level incidence rates either by gender, age, and risk class. For example, the overall trend for male cancer incidence has been relatively flat over the past several years, as can been seen in the graphs below figure 8 which are taken directly from the 2011 edition of Canadian Cancer Statistics (CCS). Of the more significant cancers, lung cancer incidence has been declining, while prostate cancer has been increasing. Since lung cancer is predominantly a disease of smokers, non-smokers are not benefiting from this trend.

Similar considerations might apply to observed trends in cardiovascular diseases as well. Observed trends at the aggregate may be misleading, reflecting instead a decrease in the prevalence of smokers.

The historic trajectory of prostate cancer incidence illustrates in a dramatic fashion how an intervention, such as the use of more sensitive diagnostic technology (i.e., the prostate-specific antigen (PSA) test), can suddenly shift the level of incidence rates and cause confusion with respect to the projection of prospective incidence rates. The PSA test was developed in the United States and was approved by the U.S. Food and Drug Administration in 1987 for cancer patients and subsequently for diagnostic screening in 1992. Canadian physicians began using the PSA test in 1990.

Up until 1990, the persistent and steady rise in prostate cancer incidence in the preceding two decades was attributed to increased detection resulting from the use of transurethral resection of the prostate in the treatment of suspected benign prostatic disease.


137

Projecting the subsequent avalanche of prostate cancer diagnoses resulting from PSA testing and the eventual decline and resumption of prostate cancer incidence trend would have been virtually impossible without a deep understanding of the underlying disease and the evolution of testing practice and physician-patient behavior.

The following table shows the evolution of actual versus projected age-standardized incidence rate for prostate cancer by year of publication of the CCS and up to the year that actual data was available (generally a five-year lag). It was not until the 1998 edition of the CCS that projections began to track future actual incidence rates more closely.

Table 6: Actual versus Projected prostate cancer incidence. Canadian Cancer Statistics, 1995 through to 2001.

The preceding example shows the challenge in forecasting future trend rates, especially when technology-driven interventions can cause dramatic shifts in diagnostic practice and hence reported incidence rates.

There are other complexities as well. As shown in the graphs in figure 8 below for cancer, trend rates vary by the specific type of cancer. Examination of data reveals that trend rates vary by age, gender, smoking habits, and province. It is therefore good practice to examine trends by attained age, gender, specific disease (by major cancer site in the case of cancer) and also to be aware of current diagnostic practices including how they may change over time.

A review of trends in insurance data is also important where it is available. Trends observed in insurance data also reflect factors such as changes in the applicant pool mix and underwriting standards and quality.

Year of Last Year Actual vs Projected Incidence Rates for Prostate Cancer (rounded)CCS of Data 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

1995 1990 100 100 103 105 108 1101996 1991 113 109 112 115 118 1241997 1992 124 120 125 128 133 1371998 1993 138 - - - - 1121999 1994 127 107 109 111 113 1142000 1995 110 108 110 113 115 1162001 1996 110 110 111 113 115 118


138


139


140

Figure 8: Canadian Cancer Statistics 2011 (Canadian Cancer Society’s Steering Committee on Cancer Statistics. Canadian Cancer Statistics 2011. Toronto, ON: Canadian Cancer Society, 2011).

b) Class Segmentation

Class segmentation adjustments refer to adjustments to the base incidence rates, which are typically only segmented by sex and age, to correspond to the premium rate classes. The discussion is limited to considerations involved in segmenting aggregate incidence rates into incidence rates for non-smoker and smokers.

The correlation between smoking and overall or cause-specific mortality has been studied and reported on extensively in scientific and insurance literature. Less common are studies that have investigated the impact of smoking on the incidence of specific illnesses, such as cancer, stroke, and heart attack.

If relative risk ratios between smokers and non-smokers for each major critical illness are available, then the aggregate incidence rates could be segmented using this information and


141

information on the prevalence of non-smokers in the population underlying the relative risk ratios. In carrying out this kind of exercise it may be useful to group the critical illnesses into general broad categories such as:

• Cancer: given the significant relationship between smoking and lung cancer incidence, it is very helpful to perform this adjustment for the major cancers separately (lung and related cancers, breast cancer, colorectal cancer, prostate cancer, and other cancers);

• Cardiovascular Diseases: heart attack, stroke, and heart-related procedures; • Neurological Diseases: Alzheimer’s, Parkinson’s, multiple sclerosis; • Accident-Related Diseases: paralysis, blindness, deafness; and • Functional Disability: loss of independent existence.

When using this kind of technique to generate smoker-distinct incidence rates, the actuary should keep in mind that the results are highly sensitive to the assumed underlying proportion of smokers.

Smoking impacts different diseases differently and also varies by age and gender. For example, in a UK general population study (BMJ 1995; 311:471) examining the impact of smoking on heart attack incidence, smokers aged 30–49 were more than five times as likely to have a non-fatal heart attack relative to non-smokers and three times as likely at ages 50–59. For some impairments, smoking appears to have no significant impact and for others, such as Parkinson’s disease, it appears to have a positive impact (Ann Neurol., December 2001; 50(6):780-6.)

It is especially useful to examine UK insured experience when estimating the risk ratios between smokers and non-smokers. The Institute and Faculty of Actuaries Continuous Mortality Investigation (CMI) Working Paper 52, Cause-specific CMI critical illness diagnosis rates for accelerated business, 2003–2006, published June 2011, is referenced. This study measured cause-specific claim diagnosis rates for accelerated CI on a “by lives/number” basis, using nearly 16,000 claims settled over 2003–2006. Despite the large number of claims, there were very few claims under age 30 or above age 60, and cause-specific rates by smoking status were only derived for causes that had a sufficient number of claims.

The two graphs (figures 9a, 9b) below show the smoker to non-smoker risk ratios by attained age, gender, and by cause. For females, smoker to non-smoker risk ratios are only available for cancer, deaths, and other CIs.

http://www.ncbi.nlm.nih.gov/pubmed/11761476�


142

Figure 9a. From CMI Working Paper 52. By number/lives basis. UK Acceleration business, policy years 5+. HA is heart attack.

Figure 9b. From CMI Working Paper 52. By number/lives basis. UK Acceleration business, policy years 5+.

Under accelerated CI products, both mortality and CI are covered. Death claims typically represent causes of death that are not related to CI. Note that in practice, a claimant who

0% 50%

100% 150% 200% 250% 300% 350% 400% 450% 500%

30 35 40 45 50 55 60

UK CI Smoker/NonSmoker Risk Ratio Males

Cancer HA OtherCI Death

0%

50%

100%

150%

200%

250%

300%

350%

30 35 40 45 50 55 60

UK CI Smoker/NonSmoker Risk Ratio Females

Cancer OtherCI Death


143

suffers a fatal heart attack would likely be classified as a death claim rather than as a CI heart attack claim.

For females, most of the claims were for cancer, with deaths and other CI splitting the difference. The proportion of total claims related to death in this study was only about 11% for non-smokers and 22% for smokers.

The smoker risk ratios for cancer are relatively modest for both males and females and near 1 at the younger ages. The female cancer risk ratios under age 40 are reported to be less than 1 which is unusual. If the true ratio is about one, then the reported results could just be due to statistical fluctuation.

In the UK, biochemical verifications (e.g., cotinine to verify smoking declaration) occur at high amounts relative to Canada. The non-smoker pool likely contains a non-trivial proportion of smokers, thereby reducing the smoker risk ratios.

Since smoker-distinct, cause-specific incidence rates for female lives were limited to the three causes shown, male data is presented on a similar basis as females in figure 10.

Figure 10: Male lives, UK smoker risk ratios using same causes as in figure 9b.

On this basis, the smoker risk ratios between males and females are reasonably consistent with each other except for cause death.

Estimating smoker-distinct incidence requires consideration of several data sources as well as consideration of the quality of underwriting and the applicant pool.

c) Risk Selection and Underwriting Risk selection and underwriting adjustments refer to the adjustments to the population-based incidence rates that reflect the impact of individual underwriting and the insurance application process.

A framework that is helpful in describing the effects of underwriting on incidence is the parametric model used to graduate the data underlying the CIA 86–92 and the CIA 97–04 tables select tables. The model can be depicted mathematically as follows:

0% 50%

100% 150% 200% 250% 300% 350%

30 35 40 45 50 55 60

UK CI Smoker/NonSmoker Risk Ratio Males-same causes as Females

Cancer OtherCI Death


144

𝑞[𝑥−𝑡]+𝑡𝑆𝑒𝑙𝑒𝑐𝑡 = 𝑞𝑥𝑈𝑙𝑡(𝑟(𝑥) + 𝐺(𝑡)(1− 𝑟(𝑥)) and,

G(t)= {(𝑡/𝛽)α for 0≤t≤β, 1 for t>β}

Where

r(x) =the ratio between select mortality (incidence) at attained age x and ultimate mortality (incidence) at attained age x. This reflects the immediate impact of underwriting and selection.

t=policy year,

β= the number of policy years that underwriting selection has an effect (i.e., the select period).

α=the parameter which defines the rate at which select mortality (incidence) approaches ultimate mortality (incidence).

Figure 11 below illustrates how selection diminishes under this simplified model assuming r(x)=0.5, β=10 for all ages, and α taking on values 0+Δε, 0.2, 0.4,0.6, 0.8, and 1.0.

Figure 11: Impact of decay parameter α with r(x)=0.5 and 10-year selection period.

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1 2 3 4 5 6 7 8 9 10

Effect of α

0% 20% 40% 60% 80% 100%


145

When α is close to 0 (i.e. 0+Δε), selection wears off immediately. When α=1, selection wears off uniformly.

For example, the parameter values used in the construction of the CIA 97–04 select mortality tables were:

Male r(x) = 0.51 at age 25 and declining slowly to a value of 0.31 at age 65.

Female r(x)=0.50 at age 25 and declining slowly to a value of 0.25 at age 65.

The select period, β, was assumed to be 15 years.

The values of α (rounded) were as follows:

Males Females Smokers (age last) 0.28 0.22 Non-Smokers (age last) 0.86 0.74

Table 7: Values of α for CIA9704 tables.

Thus, the key assumptions required under this parametric model of selection are:

a) What are the ultimate level of incidence rates, 𝑞𝑥𝑈𝑙𝑡, currently and in the absence of trends?

b) What is (are) the length of the select period, β?

c) What is the relation of the first-year incidence to the ultimate incidence; that is r(x)

= 𝑞[𝑥]𝑆𝑒𝑙𝑒𝑐𝑡

𝑞𝑥𝑈𝑙𝑡 ?

d) Finally, what is (are) the rate at which selection wears off, α?

With respect to the ultimate level of incidence rates, it will take many years of Canadian insured experience before this can be estimated with confidence. Until then, indirect approaches could be used. One such way of estimating ultimate incidence rates is to apply “insured to population” factors (“I/P” in the UK) to the population-based incidence rates. I/P factors can be inferred from the relationship between ultimate life insurance mortality to population mortality (ideally limiting the comparison to those causes of death found in CI contracts such as cancer, heart attack, and stroke). I/P factors could also be inferred from ultimate insured CI incidence to population CI incidence in the UK.

With respect to the select period, β, and the decay parameter, α, it will also be many years before these parameters are estimated with confidence. In the UK, the select period is generally thought to be no more than three to five years. It is believed that cancer exhibits a one-year selection period.

Previously-mentioned Working Paper 52 examines CI incidence by cause on a “lives/numbers” basis. The authors expressed the observed experience as a graduation relative to the population based CIBT02 tables for each policy year. The tables resulted in a good fit by policy year and provided validation with respect to their assumptions about selection.

As shown below in tables 8 and 9, the resulting table, expressed as a percentage of the CIBT02 incidence rates for cancer, has a one-year selection (β) period.


146

Table 8: Working Paper 52 best fit table for male non-smoker cancer incidence.

Table 9: Working Paper 52 best fit table for female non-smoker cancer incidence.

For example, for female non-smokers, attained age 45, morbidity at duration is 54% of the CIBT02 female cancer table at duration 0 and is 67% at subsequent durations. In addition, the value of r(x) would be 54%/67% =80.6%. Estimates of UK I/P factors for cancer are those factors in column “5+”.

Life actuaries in Canada are accustomed to long selection periods as evidenced in life insurance tables. A short selection period could be viewed with some skepticism.

While the selection period for critical illness insurance may be longer than five years, what is more important is the rate of decay of selection. If the decay to ultimate is rapid. then it makes little practical difference if the true select period is 10 years or five years, as the graph in figure 12 below illustrates. It would also be difficult to statistically differentiate these two patterns in practice.

Male non-smokers: Cancer

Age exact Curtate duration at diagnosisatdiagnosis 0 1 2 3 4 5+

30 56% 70% 70% 70% 70% 73%35 53% 67% 67% 67% 67% 69%40 50% 63% 63% 63% 63% 65%45 42% 52% 52% 52% 52% 54%50 40% 50% 50% 50% 50% 52%55 46% 58% 58% 58% 58% 60%60 44% 56% 56% 56% 56% 58%

Female non-smokers: Cancer


30 56% 71% 71% 71% 71% 71%35 59% 74% 74% 74% 74% 74%40 58% 73% 73% 73% 73% 73%45 54% 67% 67% 67% 67% 67%50 49% 61% 61% 61% 61% 61%55 49% 61% 61% 61% 61% 61%60 49% 61% 61% 61% 61% 61%


147

Figure 12: Illustration showing relative importance of decay compared to length of selection period.

Finally, every set of underwriting requirements will have different theoretical values for r(x), α, β, and 𝑞𝑥

𝑈𝑙𝑡. In general, r(x), α, and 𝑞𝑥𝑈𝑙𝑡 will be higher for business that is underwritten less intensively as compared to business that is underwritten more intensively, other things being equal.

We cannot conclude this section without a few remarks about anti-selection. Anti-selection is present wherever there is choice (selection) to the applicant or life insured. A few examples of choices that may lead to anti-selection are:

• Selecting an insurer with weaker underwriting standards; • Choosing the maximum coverage amount within an underwriting cell; • Not being completely truthful on the application; • Not reporting on recent changes in health, unusual symptoms, and other

manifestations of new disease; and • The choice between continuing or terminating coverage.

Critical illness policies provide a benefit while the insured is alive. This is a powerful incentive to select against an insurer. It has been reported that in some markets where CI is sold, anti-selection results in morbidity that is higher in the first policy year than in subsequent years. CMI Working Paper 52 reported on this phenomenon with respect to the experience in heart attack incidence for male non-smokers. This is surprising since the screening of cardiovascular risks is core to the underwriting of lives. This could be the result of statistically non-significant random variation due to a limited number of claims (80

0,000

0,200

0,400

0,600

0,800

1,000

1,200

1 2 3 4 5 6 7 8 9 10 11

Selection Period vs. Rate of Decay

(α=0.22,β=5) (α=0.40,β=5)


148

claims) or due to anti-selection from non-disclosure of risk factors related to high non-medical limits.

Table 10: Working Paper 52 best fit table for male non-smoker heart attack incidence.

d) Product Design Similar to mortality-based products, product design likely has an important influence on morbidity levels. Product design influences and sometimes forces certain choices from the insured over others. These choices can lead to anti-selection.

Product design elements that encourage high lapses should have worse morbidity when lapses are elevated, other things being equal, compared to a product design that does not have such elements. A renewable term 10 product design, with its characteristic high premium jumps every 10 years and associated shock lapses, is a familiar example. Reflection of morbidity deterioration following and during a sustained period of high lapses is necessary.

Many CI products are issued with very rich return of premium (RoP) riders which return a substantial portion of, if not all, the cumulative policy premium paid by the policyholder from inception if the policy is surrendered after 15 years. Applicants who choose this kind of product, compared to applicants who do not, might be expected to exhibit less anti-selection since total premiums are considerably higher and there is an incentive to persist with the coverage to at least the policy year when the surrender benefits commence. These considerations suggest that morbidity may be lower than average until the policy year when surrender benefits commence, and then higher than average when lapses are elevated. It should be noted that if aggregate morbidity is bifurcated so that a morbidity credit is given to such products, then a morbidity debit should be applied to the rest of the products.

Concluding Remarks Adjustments to transform population-based incidence rates to insured incidence rates are highly subjective and speculative in nature. The forthcoming CIA experience studies should contribute to our understanding of emerging CI insured experience and remove some of the ambiguity that surrounds insured incidence rates. However, it will be many years before there are sufficient claims to reliably estimate incidence rates at older ages, to determine the transition from select to ultimate, and to quantify differences due to underwriting requirements, product design, and other factors with a high degree of confidence. Until such a time, it will be important to consider the experience from other parts of the world.

Male non-smokers: Heart Attack


30 19% 15% 15% 15% 15% 17%35 19% 15% 15% 15% 15% 17%40 20% 15% 15% 15% 15% 17%45 23% 17% 17% 17% 17% 20%50 28% 21% 21% 21% 21% 24%55 31% 24% 24% 24% 24% 27%60 29% 23% 23% 23% 23% 26%


149

A future project for the Living Benefits Subcommittee could be the development of an insured table which takes into account Canadian insured experience and incorporates relevant experience and knowledge from other parts of the world.

Documents

Research Paper: Critically Canadian: Canadian Critical ... · 1990s. Canadian CI sales developed slowly at first but then experienced double-digit growth in the early 2000s. While